siunitx.tex

\iffalse meta-comment

File: siunitx.tex Copyright (C) 2014-2024 Joseph Wright

It may be distributed and/or modified under the conditions of the
LaTeX Project Public License (LPPL), either version 1.3c of this
license or (at your option) any later version.  The latest version
of this license is in the file

   https://www.latex-project.org/lppl.txt

This file is part of the "siunitx bundle" (The Work in LPPL)
and all files in that bundle must be distributed together.

The released version of this bundle is available from CTAN.

-----------------------------------------------------------------------

The development version of the bundle can be found at

   https://github.com/josephwright/siunitx

for those people who are interested.

-----------------------------------------------------------------------

\fi

\documentclass{l3doc}

% The next line is needed so that \GetFileInfo will be able to pick up
% version data (quite apart from making the demos work).
\usepackage{siunitx}
\DeclareSIUnit\noop{\relax} % For printing prefixes
\DeclareSIPower\quartic\tothefourth{4} % For demos
\DeclareSIUnit\KWH{kWh}
\DeclareSIQualifier\polymer{pol}
\DeclareSIQualifier\catalyst{cat}
\DeclareSIUnit\millimetremercury{mmHg}

% Commands for this document
\ExplSyntaxOn
\makeatletter
\NewDocumentCommand \acro { m }
  {
    \textsc
      {
        \exp_args:NV \tl_if_head_eq_charcode:nNTF \f@series { m }
          { \text_lowercase:n }
          { \use:n }
            {#1}
      }
  }
\makeatother
\ExplSyntaxOff
\NewDocumentCommand\email{m}{\href{mailto:#1}{\nolinkurl{#1}}}
\NewDocumentCommand\ext{m}{\texttt{.#1}}
\NewDocumentCommand\foreign{m}{\textit{#1}}
\NewDocumentCommand\opt{m}{\texttt{#1}}
\NewDocumentCommand\version{m}{\ensuremath{#1}}
\ExpandArgs{c}\NewDocumentCommand{version (pdfstring context)}{m}{#1}
\pdfstringdefDisableCommands{%
  \ExpandArgs{Nc}\DeclareCommandCopy\version{version (pdfstring context)}}
% Tidy up the above in bookmarks
\makeatletter
\pdfstringdefDisableCommands{%
  \let\acro\@firstofone
  \let\ext\@firstofone
  \let\foreign\@firstofone
  \let\opt\@firstofone
}
\makeatother

\NewDocumentCommand\DescribePrefix{m}{%
  #1 &
  \cs{#1} &
  \unit{\csname #1\endcsname\noop}
}
\NewDocumentCommand\DescribeUnit{O{#2}m}{%
  #1 &
  \cs{#2} &
  \unit{\csname #2\endcsname}
}

% For demos
\usepackage[portuguese,brazilian,catalan,french,german,italian,polish,spanish,UKenglish]{babel}
\AtBeginDocument{\shorthandoff{:<>}}
\usepackage{translations}
\usepackage{cancel}
\usepackage{collcell}
\usepackage{sansmath}
\newlength{\mylength}

% For creating code demonstrations
% This needs access to some code-level interfaces in listings
\usepackage{listings}
\makeatletter
\lst@RequireAspects{writefile}
\newsavebox\LaTeXdemo@box
\lstnewenvironment{LaTeXdemo}[1][code and example]
  {%
    \global\let\lst@intname\@empty
    \edef\LaTeXdemo@end{%
      \expandafter\noexpand\csname LaTeXdemo@@#1@end\endcsname
    }%
    \@nameuse{LaTeXdemo@@#1}%
  }
  {\LaTeXdemo@end}
\newcommand\LaTeXdemo@new[3]{%
  \@namedef{LaTeXdemo@@#1}{#2}%
  \@namedef{LaTeXdemo@@#1@end}{#3}%
}
\newcommand*\LaTeXdemo@common{%
  \setkeys{lst}
    {%
       basicstyle   = \small\ttfamily,
       basewidth    = 0.51em,
       gobble       = 2,
       language     = [LaTeX]{TeX},
    }%
}
\newcount\LaTeXdemo@count
\newcommand*\LaTeXdemo@input{%
  \catcode`\^^M = 10\relax
  \input{\jobname-\number\LaTeXdemo@count.tmp}%
}
\LaTeXdemo@new{code and example}{%
  \setbox\LaTeXdemo@box=\hbox\bgroup
    \global\advance\LaTeXdemo@count by 1 %
    \lst@BeginAlsoWriteFile{\jobname-\number\LaTeXdemo@count.tmp}%
    \LaTeXdemo@common
}{%
    \lst@EndWriteFile
  \egroup
  \begin{center}
    \ifdim\wd\LaTeXdemo@box > 0.48\linewidth
      \begin{minipage}{\linewidth}
        \usebox\LaTeXdemo@box
      \end{minipage}%
      \par
      \begin{minipage}{\linewidth}
        \LaTeXdemo@input
      \end{minipage}
    \else
      \begin{minipage}{0.48\linewidth}
        \LaTeXdemo@input
      \end{minipage}%
      \hspace{\fill}%
      \begin{minipage}{0.48\linewidth}
        \usebox\LaTeXdemo@box
      \end{minipage}%
    \fi
  \end{center}
}
\LaTeXdemo@new{code and float}{%
  \global\advance\LaTeXdemo@count by 1 %
  \lst@BeginAlsoWriteFile{\jobname-\number\LaTeXdemo@count.tmp}%
  \LaTeXdemo@common
}{%
  \lst@EndWriteFile
  \LaTeXdemo@input
}
\LaTeXdemo@new{code only}{\LaTeXdemo@common}{}
\makeatother

% For table demos
\usepackage{datatool}
\usepackage{multirow}
\usepackage[table]{xcolor}
% Has to come after xcolor
\usepackage{pgfplots}
\pgfplotsset{compat = 1.18, compat/show suggested version = false}

% For the extra-long tables
\usepackage{xtab}

% For the table notes
\usepackage{threeparttable}

% Design changes
\usepackage{caption}
\makeatletter
\edef\@floatboxreset{%
  \unexpanded\expandafter{\@floatboxreset}%
  \noexpand\centering
}
\makeatother
\usepackage[osf]{mathpazo}

\hypersetup{hidelinks}

\begin{document}

\GetFileInfo{siunitx.sty}

\title{%
  \pkg{siunitx} -- A comprehensive (\acro{SI}) units package%
  \thanks{This file describes \fileversion,
    last revised \filedate.}%
}

\author{%
  Joseph Wright%
  \thanks{%
    E-mail:
    \email{joseph@texdev.net}%
  }%
}

\date{Released \filedate}

\maketitle

\tableofcontents

\begin{abstract}
  Physical quantities have both numbers and units, and each physical quantity
  should be expressed as the product of a number and a unit. Typesetting
  physical quantities requires care to ensure that the combined mathematical
  meaning of the number--unit combination is clear. In particular, the
  \acro{SI} units system lays down a consistent set of units with rules on how
  these are to be used. However, different countries and publishers have
  differing conventions on the exact appearance of numbers (and units). The
  \pkg{siunitx} package provides a set of tools for authors to typeset
  quantities in a consistent way. The package has an extended set of
  configuration options which make it possible to follow varying typographic
  conventions with the same input syntax. The package includes automated
  processing of numbers and units, and the ability to control tabular alignment
  of numbers.
\end{abstract}

\begin{documentation}

\section{Introduction}

The correct application of units of measurement is very important in technical
applications. For this reason, carefully-crafted definitions of a coherent
units system have been laid down by the \foreign{Conférence Génrale des Poids
et Mesures} (\acro{CGPM}): this has resulted in the \foreign{Système
International d'Unités}~(\acro{SI}). At the same time, typographic conventions
for correctly displaying both numbers and units exist to ensure that no loss of
meaning occurs in printed matter.

The \pkg{siunitx} package aims to provide a unified method for \LaTeX{} users
to typeset numbers and units correctly and easily. The design philosophy of
\pkg{siunitx} is to follow the agreed rules by default, but to allow variation
through option settings. In this way, users can use \pkg{siunitx} to follow the
requirements of publishers, co-authors, universities, \foreign{etc}.\ without
needing to alter the input at all.

\section{\pkg{siunitx} for the impatient}

The package provides the user macros:
\begin{itemize}
  \item \cs{ang}\oarg{options}\marg{angle}
  \item \cs{num}\oarg{options}\marg{number}
  \item \cs{unit}\oarg{options}\marg{unit}
  \item \cs{qty}\oarg{options}\marg{number}\marg{unit}
  \item \cs{numlist}\oarg{options}\marg{numbers}
  \item \cs{numproduct}\oarg{options}\marg{numbers}
  \item \cs{numrange}\oarg{options}\marg{numbers}\marg{number2}
  \item \cs{qtylist}\oarg{options}\marg{numbers}\marg{unit}
  \item \cs{qtyproduct}\oarg{options}\marg{numbers}\marg{unit}
  \item \cs{qtyrange}\oarg{options}\marg{number1}\marg{number2}\marg{unit}
  \item \cs{complexnum}\oarg{options}\marg{number}
  \item \cs{complexqty}\oarg{options}\marg{number}\marg{unit}
  \item \cs{sisetup}\marg{options}
  \item \cs{tablenum}\oarg{options}\marg{number}
\end{itemize}
plus the \texttt{S} column type for decimal alignments and units in tabular
environments. These user macros and column types are designed for typesetting
numbers and units with control of appearance and with intelligent processing.

Numbers are processed with understanding of exponents, or using additional
commands for products and complex numbers.
\begin{LaTeXdemo}
  \num{12345,67890}  \\
  \num{.3e45}        \\
  \complexnum{1+-2i} \\
  \numproduct{1.654 x 2.34 x 3.430}
\end{LaTeXdemo}

The unit system can interpret units given as text to be used directly or as
macro-based units. In the latter case, different formatting is possible.
\begin{LaTeXdemo}
  \unit{kg.m.s^{-1}}                \\
  \unit{\kilogram\metre\per\second} \\
  \unit[per-mode = symbol]
    {\kilogram\metre\per\second}  \\
  \unit[per-mode = symbol]
    {\kilogram\metre\per\ampere\per\second}
\end{LaTeXdemo}

Simple lists and ranges of numbers can be handled.
\begin{LaTeXdemo}
  \numlist{10;20;30}                     \\
  \qtylist{0.13;0.67;0.80}{\milli\metre} \\
  \numrange{10}{20}                      \\
  \qtyrange{0.13}{0.67}{\milli\metre}
\end{LaTeXdemo}

A wide range of options are available to control the behavior of the package.
For example, with the standard settings all text is typeset in the current
upright math font. This can be adjusted to use text mode, follow various
aspects of the surrounding formatting, \foreign{etc}. Similarly, the standard
settings are based around an English-speaking locale, but can be adjusted
to follow the traditions of other areas.

\section{Using the \pkg{siunitx} package}

\subsection{Numbers}

\begin{function}{\num}
  \begin{syntax}
    \cs{num}\oarg{options}\marg{number}
  \end{syntax}
\end{function}
Numbers are automatically formatted by the \cs{num} macro. This takes one
optional argument, \meta{options}, and one mandatory one, \meta{number}. The
contents of \meta{number} are automatically formatted. The formatter removes
both \enquote{soft} (\verb*| |) and \enquote{hard} spaces (|\,| and |~|),
automatically identifies exponents (as standard marked using |e|, |E|, |d| or
|D|) and adds the appropriate spacing of large numbers. If required,
a leading zero is added before a decimal marker: both |.|
and "," are recognised as decimal markers.
\begin{LaTeXdemo}
  \num{123}     \\
  \num{1234}    \\
  \num{12345}   \\
  \num{0.123}   \\
  \num{0,1234}  \\
  \num{.12345}  \\
  \num{3.45d-4} \\
  \num{-e10}
\end{LaTeXdemo}

Note that numbers are parsed before typesetting, which does have a performance
overhead (only obvious with very large amounts of numerical input). The parser
understands a range of input syntaxes, as demonstrated above.

\begin{function}{\numlist}
  \begin{syntax}
    \cs{numlist}\oarg{options}\marg{numbers}
  \end{syntax}
\end{function}
Lists of numbers may be processed using the \cs{numlist} function. Each
\meta{number} is given within the list of \meta{numbers} within a brace pair,
as the list can have a flexible length.
\begin{LaTeXdemo}
  \numlist{10;30;50;70}
\end{LaTeXdemo}

\begin{function}{\numproduct}
  \begin{syntax}
    \cs{numproduct}\oarg{options}\marg{numbers}
  \end{syntax}
\end{function}
Runs of products of numbers may be inserted using the \cs{numproduct} function.
This acts in the same way as \cs{num}, but inserts either a symbol or phrase
between the entries. The latter should be separated by |x| tokens.
\begin{LaTeXdemo}
  \numproduct{10 x 30}
\end{LaTeXdemo}

\begin{function}{\numrange}
  \begin{syntax}
    \cs{numrange}\oarg{options}\marg{number1}\marg{number2}
  \end{syntax}
\end{function}
Simple ranges of numbers can be handled using the \cs{numrange} function. This
acts in the same way as \cs{num}, but inserts a phrase or other text between
the two entries.
\begin{LaTeXdemo}
  \numrange{10}{30}
\end{LaTeXdemo}

\subsection{Angles}

\begin{function}{\ang}
  \begin{syntax}
    \cs{ang}\oarg{options}\marg{angle}
  \end{syntax}
\end{function}
Angles can be typeset using the \cs{ang} command. The \meta{angle} can be given
either as a decimal number or as a semi-colon separated list of degrees,
minutes and seconds, which is called \enquote{arc format} in this document. The
numbers which make up an angle are processed using the same system as other
numbers.
\begin{LaTeXdemo}
  \ang{10}    \\
  \ang{12.3}  \\
  \ang{4,5}   \\
  \ang{1;2;3} \\
  \ang{;;1}   \\
  \ang{+10;;} \\
  \ang{-0;1;}
\end{LaTeXdemo}

\subsection{Units}

\begin{function}{\unit}
  \begin{syntax}
    \cs{unit}\oarg{options}\marg{unit}
  \end{syntax}
\end{function}
The symbol for a unit can be typeset using the \cs{unit} macro: this provides
full control over output format for the unit. Like the \cs{num} macro,
\cs{unit} takes one optional and one mandatory argument. The unit formatting
system can accept two types of input. When the \meta{unit} contains literal
items (for example letters or numbers) then \pkg{siunitx} converts |.| and |~|
into inter-unit product and correctly positions sub- and superscripts specified
using |_| and |^|. The formatting methods will work with both math and text
mode.
\begin{LaTeXdemo}
  \unit{kg.m/s^2} \\
  \unit{g_{polymer}~mol_{cat}.s^{-1}}
\end{LaTeXdemo}

The second operation mode for the \cs{unit} macro is an \enquote{interpreted}
system. Here each unit, \acro{SI} multiple prefix and power is given a macro
name. These are entered in a method very similar to the reading of the unit
name in English.
\begin{LaTeXdemo}
  \unit{\kilo\gram\metre\per\square\second} \\
  \unit{\gram\per\cubic\centi\metre}        \\
  \unit{\square\volt\cubic\lumen\per\farad} \\
  \unit{\metre\squared\per\gray\cubic\lux}  \\
  \unit{\henry\second}
\end{LaTeXdemo}

On its own, this is less convenient than the direct method, although it does
use meaning rather than appearance for input. However, the package allows you
to define new unit macros; a large number of pre-defined abbreviations are also
supplied. More importantly, by defining macros for units, instead of literal
input, new functionality is made available. By altering the settings used by
the package, the same input can yield a variety of different output formats.
For example, the \cs{per} macro can give reciprocal powers, slashes or be used
to construct units as fractions.

\begin{function}{\qty}
  \begin{syntax}
    \cs{qty}\oarg{options}\marg{number}\marg{unit}
  \end{syntax}
\end{function}
Very often, numbers and units are given together. Formally, the value of a
quantity is the product of the number and the unit, the space being regarded as
a multiplication sign. The \cs{qty} macro combines the
functionality of \cs{num} and \cs{unit}, and makes this both possible and easy.
The \meta{number} and \meta{unit} arguments work exactly like those for the
\cs{num} and \cs{unit} macros, respectively.
\begin{LaTeXdemo}
  \qty[mode = text]{1.23}{J.mol^{-1}.K^{-1}}          \\
  \qty{.23e7}{\candela}                               \\
  \qty[per-mode = symbol]{1.99}{\per\kilogram}        \\
  \qty[per-mode = fraction]{1,345}{\coulomb\per\mole}
\end{LaTeXdemo}

It is possible to set up the unit macros to be available outside of the
\cs{qty} and \cs{unit} functions. This is not the standard behaviour as there
is the risk of name clashes (for example, \cs{day} is a \TeX{} primitive
and several packages define \cs{degree}). Full details of using \enquote{stand
alone} units are found in \ref{sec:units:creating}.

\begin{function}{\qtylist}
  \begin{syntax}
    \cs{qtylist}\oarg{options}\marg{numbers}\marg{unit}
  \end{syntax}
\end{function}
Lists of numbers with units can be handled using the \cs{qtylist} function. The
behaviour of this function is similar to \cs{numlist}, but with the addition of
the unit to each number.
\begin{LaTeXdemo}
  \qtylist{10;30;45}{\metre}
\end{LaTeXdemo}

\begin{function}{\qtyproduct}
  \begin{syntax}
    \cs{qtyproduct}\oarg{options}\marg{numbers}\marg{unit}
  \end{syntax}
\end{function}
Runs of products of numbers with units can be handled using the
\cs{qtyproduct} function. The behaviour of this function is similar to
\cs{numproduct}, but with the addition of a unit to each number.
\begin{LaTeXdemo}
  \qtyproduct{10 x 30 x 45}{\metre}
\end{LaTeXdemo}

\begin{function}{\qtyrange}
  \begin{syntax}
    \cs{qtyrange}\oarg{options}\marg{number1}\marg{number2}\marg{unit}
  \end{syntax}
\end{function}
Ranges of numbers with units can be handled using the \cs{qtyrange}
function. The behaviour of this function is similar to \cs{numrange}, but
with the addition of a unit to each number.
\begin{LaTeXdemo}
  \qtyrange{10}{30}{\metre}
\end{LaTeXdemo}

The input of lists, products and ranges of quantities using a single command
allows them to be adjusted together. These commands are intended to allow
consistent formatting of related values: as such, they apply a single
unit to all of the values. This is particularly notable when using
adjustment of the numerical values.

\subsection{Complex numbers and quantities}

\begin{function}{\complexnum}
  \begin{syntax}
    \cs{complexnum}\oarg{options}\marg{number}
  \end{syntax}
\end{function}
Typesets the complex number, which can be given in the Cartesian form $a +
b\mathrm{i}$ or $a + \mathrm{i}b$, or in the polar form $r$\texttt{:}$\theta$.
Processing of the numerical parts is otherwise identical to the standard
\cs{num} command.

\begin{function}{\complexqty}
  \begin{syntax}
    \cs{complexqty}\oarg{options}\marg{number}\marg{unit}
  \end{syntax}
\end{function}
Typesets the complex number, which can be given in the Cartesian form $a +
b\mathrm{i}$ or $a + \mathrm{i}b$, or in the polar form $r$\texttt{:}$\theta$.
Processing of the numerical parts is otherwise identical to the standard
\cs{qty} command.

\subsection{The unit macros}

The package always defines the basic set of \acro{SI} units with macro names.
This includes the base \acro{SI} units, the derived units with special names
and the prefixes. A small number of powers are also given pre-defined names.
Full details of units in the \acro{SI} are available on-line~\cite{BIPM}.

The seven base \acro{SI} units are always defined (Table~\ref{tab:unit:base}).
In addition, the macro \cs{meter} is available as an alias for \cs{metre}, for
users of US spellings. The full details of the base units are given in the
\acro{SI} Brochure.
\begin{table}
  \caption{\acro{SI} base units.%
    \label{tab:unit:base}}
  \begin{tabular}{@{}lll@{}}
    \toprule
      Unit & Command & Symbol \\
    \midrule
      \DescribeUnit{ampere}   \\
      \DescribeUnit{candela}  \\
      \DescribeUnit{kelvin}   \\
      \DescribeUnit{kilogram} \\
      \DescribeUnit{metre}    \\
      \DescribeUnit{mole}     \\
      \DescribeUnit{second}   \\
    \bottomrule
  \end{tabular}
\end{table}

The \acro{SI} also lists a number of units which have special names and
symbols: these are listed in Table~\ref{tab:unit:derived}.
\begin{table}
  \caption{Coherent derived units in the \acro{SI} with special names and
      symbols.%
      \label{tab:unit:derived}}
  \begin{tabular}{@{}lll>{\qquad}lll@{}}
    \toprule
      Unit & Command & Symbol & Unit & Command & Symbol \\
    \midrule
      \DescribeUnit{becquerel} &
      \DescribeUnit{newton}    \\
      \DescribeUnit[degree Celsius]{degreeCelsius} &
      \DescribeUnit{ohm}       \\
      \DescribeUnit{coulomb}   &
      \DescribeUnit{pascal}    \\
      \DescribeUnit{farad}     &
      \DescribeUnit{radian}    \\
      \DescribeUnit{gray}      &
      \DescribeUnit{siemens}   \\
      \DescribeUnit{hertz}     &
      \DescribeUnit{sievert}   \\
      \DescribeUnit{henry}     &
      \DescribeUnit{steradian} \\
      \DescribeUnit{joule}     &
      \DescribeUnit{tesla}     \\
      \DescribeUnit{lumen}     &
      \DescribeUnit{volt}      \\
      \DescribeUnit{katal}     &
      \DescribeUnit{watt}      \\
      \DescribeUnit{lux}       &
      \DescribeUnit{weber}     \\
    \bottomrule
  \end{tabular}
\end{table}

In addition to the official \acro{SI} units, \pkg{siunitx} also provides macros
for a number of units which are accepted for use in the \acro{SI} although they
are not \acro{SI} units. Table~\ref{tab:unit:accepted} lists the
\enquote{accepted} units. The command \cs{percent} is also provided for use in
units: this is accepted with the \acro{SI} as detailed in Section~5.4.7 of the
Brochure.
\begin{table}
  \caption{Non-\acro{SI} units accepted for use with the International System of
    Units.%
      \label{tab:unit:accepted}}
  \begin{tabular}{@{}lll@{}}
    \toprule
      Unit & Command & Symbol \\
    \midrule
      \DescribeUnit{astronomicalunit} \\
      \DescribeUnit{bel}              \\
      \DescribeUnit{dalton}           \\
      \DescribeUnit{day}              \\
      \DescribeUnit{decibel}          \\
      \DescribeUnit{degree}           \\
      \DescribeUnit{electronvolt}     \\
      \DescribeUnit{hectare}          \\
      \DescribeUnit{hour}             \\
      \DescribeUnit{litre} \\
       & \cs{liter} & \unit{\liter} \\
      \DescribeUnit[minute (plane angle)]{arcminute} \\
      \DescribeUnit[minute (time)]{minute} \\
      \DescribeUnit[second (plane angle)]{arcsecond} \\
      \DescribeUnit{neper}            \\
      \DescribeUnit{tonne}            \\
    \bottomrule
  \end{tabular}
\end{table}

In addition to the units themselves, \pkg{siunitx} provides pre-defined macros
for all of the \acro{SI} prefixes (Table~\ref{tab:unit:prefix}).
The spelling \enquote{\cs{deka}} is provided for US users as an alternative to
\cs{deca}.
\begin{table}
  \caption{SI prefixes.%
    \label{tab:unit:prefix}}
  \sisetup{table-number-alignment = right, table-format = 2}
  \begin{tabular}{@{}llcS[table-format = -2]>{\qquad}llcS@{}}
    \toprule
      Prefix & Command & Symbol & \multicolumn{1}{l}{Power} &
      Prefix & Command & Symbol & \multicolumn{1}{l@{}}{Power} \\
    \midrule
      \DescribePrefix{quecto} & -30 &
      \DescribePrefix{deca}   &   1 \\
      \DescribePrefix{ronto}  & -27 &
      \DescribePrefix{hecto}  &   2 \\
      \DescribePrefix{yocto}  & -24 &
      \DescribePrefix{kilo}   &   3 \\
      \DescribePrefix{zepto}  & -21 &
      \DescribePrefix{mega}   &   6 \\
      \DescribePrefix{atto}   & -18 &
      \DescribePrefix{giga}   &   9 \\
      \DescribePrefix{femto}  & -15 &
      \DescribePrefix{tera}   &  12 \\
      \DescribePrefix{pico}   & -12 &
      \DescribePrefix{peta}   &  15 \\
      \DescribePrefix{nano}   &  -9 &
      \DescribePrefix{exa}    &  18 \\
      \DescribePrefix{micro}  &  -6 &
      \DescribePrefix{zetta}  &  21 \\
      \DescribePrefix{milli}  &  -3 &
      \DescribePrefix{yotta}  &  24 \\
      \DescribePrefix{centi}  &  -2 &
      \DescribePrefix{ronna}  &  27 \\
      \DescribePrefix{deci}   &  -1 &
      \DescribePrefix{quetta} &  30 \\
    \bottomrule
  \end{tabular}
\end{table}

A small number of pre-defined powers are provided as macros. \cs{square} and
\cs{cubic} are intended for use before units, with \cs{squared} and \cs{cubed}
going after the unit.
\begin{LaTeXdemo}
  \unit{\square\becquerel} \\
  \unit{\joule\squared\per\lumen} \\
  \unit{\cubic\lux\volt\tesla\cubed}
\end{LaTeXdemo}
Generic powers can be inserted on a one-off basis using the \cs{tothe} and
\cs{raiseto} macros. These are the only macros for units which take an
argument:
\begin{LaTeXdemo}
  \unit{\henry\tothe{5}} \\
  \unit{\raiseto{4.5}\radian}
\end{LaTeXdemo}
Reciprocal powers are indicated using the \cs{per} macro. This applies to the
next unit only, unless the \opt{sticky-per} option is turned on.
\begin{LaTeXdemo}
  \unit{\joule\per\mole\per\kelvin} \\
  \unit{\joule\per\mole\kelvin} \\
  \unit{\per\henry\tothe{5}} \\
  \unit{\per\square\becquerel}
\end{LaTeXdemo}
As for generic powers, generic qualifiers are also available using the \cs{of}
function:
\begin{LaTeXdemo}
  \unit{\kilogram\of{metal}} \\
  \qty[qualifier-mode = bracket]
    {0.1}{\milli\mole\of{cat}\per\kilogram\of{prod}}
\end{LaTeXdemo}

When the \pkg{cancel} package is loaded, it is possible to \enquote{cancel out}
units using the \cs{cancel} macro. This applies to the next unit, in a similar
manner to a prefix. The \cs{highlight} macro is also available to selectively
color units. Both \cs{cancel} and \cs{highlight} are outside of the normal
semantic meaning of units, but are provided as they may be useful in some
cases.
\begin{LaTeXdemo}
  \unit[per-mode = fraction]
    {\cancel\kilogram\metre\per\cancel\kilogram\per\second} \\
  \unit{\highlight{red}\kilogram\metre\per\second} \\
  \unit[unit-color = purple]
    {\highlight{blue}\kilogram\metre\per\second}
\end{LaTeXdemo}

\subsection{Unit abbreviations}

In addition to the \enquote{full} names, \pkg{siunitx} loads a set of
abbreviated versions of the \acro{SI} units (Table~\ref{tab:unit:abbr}). The
standard \pkg{siunitx} settings only create these abbreviations within the
scope of the \cs{unit} and \cs{qty} functions, meaning that no clashes should
occur (for example with the standard \cs{pm} symbol).
\begin{center}
  \tablecaption{Unit abbreviations}
  \label{tab:unit:abbr}
  \tablefirsthead{%
    \toprule
      \multicolumn{1}{@{}l}{Unit}      &
      \multicolumn{1}{l}{Abbreviation} &
      \multicolumn{1}{l@{}}{Symbol} \\
    \midrule
  }
  \tablehead{%
    \multicolumn{3}{@{}l@{}}{\emph{Continued from previous page}} \\
    \toprule
      \multicolumn{1}{@{}l}{Unit}      &
      \multicolumn{1}{l}{Abbreviation} &
      \multicolumn{1}{l@{}}{Symbol} \\
    \midrule
  }
  \tabletail{%
    \bottomrule
    \multicolumn{3}{@{}r@{}}{\emph{Continued on next page}} \\
  }
  \tablelasttail{\bottomrule}
  \begin{xtabular}{@{}lcc@{}}
      \DescribeUnit[femtogram]{fg} \\
      \DescribeUnit[picogram]{pg}  \\
      \DescribeUnit[nanogram]{ng}  \\
      \DescribeUnit[microgram]{ug} \\
      \DescribeUnit[milligram]{mg} \\
      \DescribeUnit[gram]{g}       \\
      \DescribeUnit[kilogram]{kg}  \\

      \midrule

      \DescribeUnit[picometre]{pm}  \\
      \DescribeUnit[nanometre]{nm}  \\
      \DescribeUnit[micrometre]{um} \\
      \DescribeUnit[millimetre]{mm} \\
      \DescribeUnit[centimetre]{cm} \\
      \DescribeUnit[decimetre]{dm}  \\
      \DescribeUnit[metre]{m}       \\
      \DescribeUnit[kilometre]{km}  \\

      \midrule

      \DescribeUnit[attosecond]{as}  \\
      \DescribeUnit[femtosecond]{fs} \\
      \DescribeUnit[picosecond]{ps}  \\
      \DescribeUnit[nanosecond]{ns}  \\
      \DescribeUnit[microsecond]{us} \\
      \DescribeUnit[millisecond]{ms} \\
      \DescribeUnit[second]{s}       \\

      \midrule

      \DescribeUnit[femtomole]{fmol} \\
      \DescribeUnit[picomole]{pmol}  \\
      \DescribeUnit[nanomole]{nmol}  \\
      \DescribeUnit[micromole]{umol} \\
      \DescribeUnit[millimole]{mmol} \\
      \DescribeUnit[mole]{mol}       \\
      \DescribeUnit[kilomole]{kmol}  \\

      \midrule

      \DescribeUnit[picoampere]{pA}  \\
      \DescribeUnit[nanoampere]{nA}  \\
      \DescribeUnit[microampere]{uA} \\
      \DescribeUnit[milliampere]{mA} \\
      \DescribeUnit[ampere]{A}       \\
      \DescribeUnit[kiloampere]{kA}  \\

      \midrule

      \DescribeUnit[microlitre]{ul} \\
      \DescribeUnit[millilitre]{ml} \\
      \DescribeUnit[litre]{l}       \\
      \DescribeUnit[hectolitre]{hl} \\
      \DescribeUnit[microliter]{uL} \\
      \DescribeUnit[milliliter]{mL} \\
      \DescribeUnit[liter]{L}       \\
      \DescribeUnit[hectoliter]{hL} \\

      \midrule

      \DescribeUnit[millihertz]{mHz} \\
      \DescribeUnit[hertz]{Hz}       \\
      \DescribeUnit[kilohertz]{kHz}  \\
      \DescribeUnit[megahertz]{MHz}  \\
      \DescribeUnit[gigahertz]{GHz}  \\
      \DescribeUnit[terahertz]{THz}  \\

      \midrule

      \DescribeUnit[millinewton]{mN} \\
      \DescribeUnit[newton]{N}       \\
      \DescribeUnit[kilonewton]{kN}  \\
      \DescribeUnit[meganewton]{MN}  \\

      \midrule

      \DescribeUnit[pascal]{Pa}      \\
      \DescribeUnit[kilopascal]{kPa} \\
      \DescribeUnit[megapacal]{MPa}  \\
      \DescribeUnit[gigapascal]{GPa} \\

      \midrule

      \DescribeUnit[milliohm]{mohm} \\
      \DescribeUnit[kilohm]{kohm}   \\
      \DescribeUnit[megohm]{Mohm}   \\

      \midrule

      \DescribeUnit[picovolt]{pV}  \\
      \DescribeUnit[nanovolt]{nV}  \\
      \DescribeUnit[microvolt]{uV} \\
      \DescribeUnit[millivolt]{mV} \\
      \DescribeUnit[volt]{V}       \\
      \DescribeUnit[kilovolt]{kV}  \\

      \midrule

      \DescribeUnit[watt]{W}                \\
      \DescribeUnit[nanowatt]{nW}           \\
      \DescribeUnit[microwatt]{uW}          \\
      \DescribeUnit[milliwatt]{mW}          \\
      \DescribeUnit[kilowatt]{kW}           \\
      \DescribeUnit[megawatt]{MW}           \\
      \DescribeUnit[gigawatt]{GW}           \\
      \DescribeUnit[joule]{J}               \\
      \DescribeUnit[microjoule]{uJ}         \\
      \DescribeUnit[millijoule]{mJ}         \\
      \DescribeUnit[kilojoule]{kJ}          \\
      \DescribeUnit[electronvolt]{eV}       \\
      \DescribeUnit[millielectronvolt]{meV} \\
      \DescribeUnit[kiloelectronvolt]{keV}  \\
      \DescribeUnit[megaelectronvolt]{MeV}  \\
      \DescribeUnit[gigaelectronvolt]{GeV}  \\
      \DescribeUnit[teraelectronvolt]{TeV}  \\
      \DescribeUnit[kilowatt hour]{kWh}     \\

      \midrule

      \DescribeUnit[farad]{F}       \\
      \DescribeUnit[femtofarad]{fF} \\
      \DescribeUnit[picofarad]{pF}  \\
      \DescribeUnit[nanofarad]{nF}  \\
      \DescribeUnit[microfarad]{uF} \\
      \DescribeUnit[millifarad]{mF} \\

      \midrule

      \DescribeUnit[henry]{H}       \\
      \DescribeUnit[femtohenry]{fH} \\
      \DescribeUnit[picohenry]{pH}  \\
      \DescribeUnit[nanohenry]{nH}  \\
      \DescribeUnit[millihenry]{mH} \\
      \DescribeUnit[microhenry]{uH} \\

      \midrule

      \DescribeUnit[coulomb]{C}       \\
      \DescribeUnit[nanocoulomb]{nC}  \\
      \DescribeUnit[millicoulomb]{mC} \\
      \DescribeUnit[microcoulomb]{uC} \\

      \midrule

      \DescribeUnit[tesla]{T}       \\
      \DescribeUnit[millitesla]{mT} \\
      \DescribeUnit[microtesla]{uT} \\

      \midrule

      \DescribeUnit[kelvin]{K} \\

      \midrule

      \DescribeUnit[decibel]{dB} \\

  \end{xtabular}
\end{center}

\DescribeMacro{bit}
\DescribeMacro{byte}
Binary data is expressed in units of bits and bytes. These are normally given
prefixes which use powers of two, rather than the powers of ten used by the
\acro{SI} prefixes. As these binary prefixes are closely related to the
\acro{SI} prefixes, they are defined by \pkg{siunitx}.
\begin{table}
  \caption{Binary prefixes.%
    \label{tab:unit:binary}}
  \sisetup{table-number-alignment = right, table-format = 2}
  \begin{tabular}{@{}llcS@{}}
    \toprule
      Prefix & Command & Symbol & \multicolumn{1}{l@{}}{Power} \\ \\
    \midrule
      \DescribePrefix{kibi} & 10 \\
      \DescribePrefix{mebi} & 20 \\
      \DescribePrefix{gibi} & 30 \\
      \DescribePrefix{tebi} & 40 \\
      \DescribePrefix{pebi} & 50 \\
      \DescribePrefix{exbi} & 60 \\
      \DescribePrefix{zebi} & 70 \\
      \DescribePrefix{yobi} & 80 \\
    \bottomrule
  \end{tabular}
\end{table}

\subsection{Creating new macros}

The various macro components of a unit have to be defined before they can be
used. The package supplies a number of common definitions, but new definitions
are also possible. As the definition of a logical unit should remain the same
in a single document, these creation functions are all preamble-only.

\begin{function}{\DeclareSIUnit}
  \begin{syntax}
    \cs{DeclareSIUnit}\oarg{options}\marg{unit}\marg{symbol}
  \end{syntax}
\end{function}
New units are produced using the \cs{DeclareSIUnit} macro. The \meta{symbol}
can contain literal input, other units, multiple prefixes, powers and \cs{per},
although literal text should not be intermixed with unit macros. Units can be
created with \meta{options} from the usual list understood by \pkg{siunitx},
and apply the specific unit macro only. The (first) optional argument to
\cs{qty} and \cs{unit} can be used to override the settings for the unit: an
example is the \cs{degree} unit.
\begin{LaTeXdemo}
  \qty{3.1415}{\degree}
\end{LaTeXdemo}
 This is declared in the package (effectively) as
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit[quantity-product = {}]
    \degree{\text{\textdegree}}
\end{LaTeXdemo}
 The spacing can still be altered at point of use:
\begin{LaTeXdemo}
  \qty{67890}{\degree} \\
  \qty[quantity-product = \,]{67890}{\degree}
\end{LaTeXdemo}
The meaning of a pre-defined unit can be altered by using \cs{DeclareSIUnit}
after loading \pkg{siunitx}. This will overwrite the original definition with
the newer version.

\begin{function}{\DeclareSIPrefix}
  \begin{syntax}
    \cs{DeclareSIPrefix}\marg{prefix}\marg{symbol}\marg{powers-ten}
  \end{syntax}
\end{function}
The standard \acro{SI} powers of ten are defined by the package, and are
described above. However, the user can define new prefixes with
\cs{DeclareSIPrefix}. For example, \cs{kilo} is defined
\begin{LaTeXdemo}[code only]
  \DeclareSIPrefix\kilo{k}{3}
\end{LaTeXdemo}

\begin{function}{\DeclareSIPower}
  \begin{syntax}
    \cs{DeclareSIPower}\marg{symbol-before}\marg{symbol-after}\marg{power}
  \end{syntax}
\end{function}
This function creates two symbols, one for use before a unit, the second
for use after a unit, both of which are equivalent to the \meta{power}.
For example, one might use
\begin{LaTeXdemo}[code only]
  \DeclareSIPower\quartic\tothefourth{4}
\end{LaTeXdemo}
 with the functions then used in the document as
\begin{LaTeXdemo}
  \unit{\kilogram\tothefourth}\\
  \unit{\quartic\metre}
\end{LaTeXdemo}

\begin{function}{\DeclareSIQualifier}
  \begin{syntax}
    \cs{DeclareSIQualifier}\marg{qualifier}\marg{symbol}
  \end{syntax}
\end{function}
Following the syntax of the other macros, qualifiers may be created using the
\cs{DeclareSIQualifier} command. In contrast to the other parts of a unit,
there are no pre-defined qualifiers. It is therefore entirely up to the user to
create these. For example, to identify the mass of a product created when using
a particular catalyst, the preamble could contain:
\begin{LaTeXdemo}[code only]
  \DeclareSIQualifier\polymer{pol}
  \DeclareSIQualifier\catalyst{cat}
\end{LaTeXdemo}
 and then in the body the document could read
\begin{LaTeXdemo}
  \qty{1.234}{\gram\polymer\per\mole\catalyst\per\hour}
\end{LaTeXdemo}

\subsection{Tabular material}

Aligning numbers in tabular content is handled by a new column type, the
\texttt{S} column. This new column type can align material using a number of
different strategies, with the aim of flexibility of output without needing to
alter the input. The method used as standard is to place the decimal marker in
the number at the centre of the cell and to align the material appropriately
(Table~\ref{tab:S:standard}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Standard behaviour of the \texttt{S} column type.%
      \label{tab:S:standard}}
    \begin{tabular}{@{}S@{}}
    \toprule
      {Some Values} \\
    \midrule
         2.3456 \\
        34.2345 \\
        -6.7835 \\
        90.473  \\
      5642.5    \\
          1.2e3 \\
            e4  \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

The \texttt{S} column will attempt to automatically detect material which
should be placed before or after a number, and will maintain the alignment of
the numerical data (Table~\ref{tab:S:extras}). If the material could be
mistaken for part of a number, it should be protected by braces. The use of
\cs{color} in a table cell will also be detected and will override any general
color applied by \pkg{siunitx}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Detection of surrounding material in an \texttt{S} column.%
      \label{tab:S:extras}}
    \begin{tabular}{@{}S@{}}
    \toprule
      {Some Values} \\
    \midrule
      12.34 \\
      \color{purple} 975,31 \\
      44.268 \textsuperscript{\emph{a}} \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\begin{function}{\tablenum}
  \begin{syntax}
    \cs{tablenum}\oarg{options}\marg{number}
  \end{syntax}
\end{function}
Within more complex tables, aligned numbers may be desirable within the
argument of \cs{multicolumn} or \cs{multirow}.\footnote{Provided by the
\pkg{multirow} package} The \cs{tablenum} function is available to achieve
alignment in these situations: this is, in effect, a macro version of the
\texttt{S} column (Table~\ref{tab:tablenum}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Controlling complex alignment with the \cs{tablenum} macro.%
      \label{tab:tablenum}}
    \begin{tabular}{@{}lr@{}}
      \toprule
      Heading & Heading \\
      \midrule
      Info & More info \\
      Info & More info \\
      \multicolumn{2}{c}{\tablenum[table-format = 4.4]{12,34}}    \\
      \multicolumn{2}{c}{\tablenum[table-format = 4.4]{333.5567}} \\
      \multicolumn{2}{c}{\tablenum[table-format = 4.4]{4563.21}}  \\
      \bottomrule
    \end{tabular}
    \hspace{\fill}%
    \begin{tabular}{@{}lr@{}}
      \toprule
      Heading & Heading \\
      \midrule
      \multirow{2}*{\tablenum{88,999}} & aaa \\
                                       & bbb \\
      \multirow{2}*{\tablenum{33,435}} & ccc \\
                                       & ddd \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\section{Package control options}

\subsection{The key--value control system}

 The package uses a range of different key types:
\begin{description}
  \item[\texttt{Choice}] Takes a limited number of choices, which are described
    separately for each key.
  \item[\texttt{Integer}] Requires a number as the argument.
  \item[\texttt{Length}] Requires a length, either as a literal
    value such as \texttt{2.0cm}, or stored as a \LaTeX{} length.
  \item[\texttt{Literal}] A key which uses the value(s) given directly,
    either to check input or in output.
  \item[\texttt{Macro}] Requires a macro, which may need a single argument.
  \item[\texttt{Math}] Similar to a \texttt{literal} option, but the input is
    always used in math mode, irrespective of other \pkg{siunitx} settings.
    Thus to text-mode only input must be placed inside the argument of a
    \cs{text} macro.
  \item[\texttt{Meta}] These are options which actually apply a number of
    other options.
  \item[\texttt{Switch}] These are on--off switches, and recognise \opt{true}
    and \opt{false}.  Giving just the key name also turns the key on.
\end{description}
The tables of option names use these descriptions to indicate how the keys
should be used.

\subsection{Printing%
  \label{sec:print}}

The \pkg{siunitx} package can control the font used to print output
independently of the surrounding material. Which aspects of the font follow
those of the surroundings is influenced by a range of setting as detailed in
Table~\ref{tab:opt:print}.
\begin{table}
  \caption{Print options.%
    \label{tab:opt:print}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      color                    & Literal & \meta{none}          \\
      mode                     & Choice  & math                 \\
      number-color             & Literal & \meta{none}          \\
      number-mode              & Choice  & math                 \\
      propagate-math-font      & Switch  & false                \\
      reset-math-version       & Switch  & true                 \\
      reset-text-family        & Switch  & true                 \\
      reset-text-series        & Switch  & true                 \\
      reset-text-shape         & Switch  & true                 \\
      text-family-to-math      & Switch  & false                \\
      text-font-command        & Literal & \meta{none}          \\
      text-subscript-command   & Literal & \cs{textsubscript}   \\
      text-superscript-command & Literal & \cs{textsuperscript} \\
      text-series-to-math      & Switch  & false                \\
      unit-color               & Literal & \meta{none}          \\
      unit-mode                & Choice  & math                 \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{mode}
\DescribeOption{number-mode}
\DescribeOption{unit-mode}
The \opt{mode} option determines whether \pkg{siunitx} uses math or text mode
when printing output. The choices are \opt{match}, \opt{math}, \opt{text}. The
\opt{match} setting means that printing uses the prevailing mode unchanged
whereas \opt{math} and \opt{text} select the appropriate \TeX{} mode. It is
possible to have different fonts in math and text modes, which will highlight
the difference. The font settings which apply are also different depending on
the mode. As well as the overall setting, it is possible to apply mode
to numbers and units separately using the \opt{number-mode} and \opt{unit-mode}
options.

\DescribeOption{reset-text-family}
\DescribeOption{reset-text-series}
\DescribeOption{reset-text-shape}
When printing in text mode, the options \opt{reset-text-family},
\opt{reset-text-series} and \opt{reset-text-shape} apply. When these are
active, \pkg{siunitx} resets the relevant font selection axis property
when printing: the standard font setting is upright mid-weight roman
(|\upshape| |\mdseries| |\rmfamily|).
\begin{LaTeXdemo}
  \sisetup{mode = text}
  {\itshape  \num{1234}}\\
  {\bfseries \num{1234}}\\
  {\sffamily \num{1234}}\\
  \sisetup{
    reset-text-family = false ,
    reset-text-series = false ,
    reset-text-shape  = false
  }
  {\itshape  \num{1234}}\\
  {\bfseries \num{1234}}\\
  {\sffamily \num{1234}}\\
\end{LaTeXdemo}

\DescribeOption{propagate-math-font}
\DescribeOption{reset-math-version}
In math mode, the font used by \LaTeX{} is \enquote{invariant}, and this is
reflected in the options available. With the standard settings, in math mode
printing uses the standard math font and version (weight). The option
\opt{propagate-math-font} may be used to apply the prevailing math font
to the printed material. The setting \opt{reset-math-version} controls
whether the math version is reset or not. Note that math version is typically
used to set \enquote{bold math} but may also be used for other effects, for
example all sanserif math.
\begin{LaTeXdemo}
  {\boldmath \unit{\kilogram}}\\
  {\sansmath $\unit{\kilogram}$}\\
  {$\mathsf{\unit{\kilogram}}$}\\
  \sisetup{
    propagate-math-font = true  ,
    reset-math-version  = false
  }
  {\boldmath \unit{\kilogram}}\\
  {\sansmath $\unit{\kilogram}$}\\
  {$\mathsf{\unit{\kilogram}}$}
\end{LaTeXdemo}

\DescribeOption{text-family-to-math}
\DescribeOption{text-series-to-math}
The options \opt{text-family-to-math} and \opt{text-series-to-math} can be
used to match (as far as possible) math mode output to the surrounding text.
These options work by detecting the current text settings and making the
appropriate choice in math mode.
\begin{LaTeXdemo}
  {\sffamily \unit{\kilogram}}\\
  {\bfseries $\unit{\kilogram}$}\\
  \sisetup{
    text-family-to-math = true  ,
    text-series-to-math = true
  }
  {\sffamily \unit{\kilogram}}\\
  {\bfseries $\unit{\kilogram}$}
\end{LaTeXdemo}

\DescribeOption{text-font-command}
In some circumstances, it may be desirable to use a non-standard font command
when printing in text mode. This might be used for example to switch from
old-style to lining numbers whilst still using text mode. This may be achieved
by setting \opt{text-font-command}. For example, this document uses old-style
numbers in text mode as-standard, which can be over-ridden by selecting the
font variant which does not feature them.
\begin{LaTeXdemo}
  \sisetup{number-mode = text}
  \qty{123456789}{\kilo\volt\per\centi\metre} \\
  \sisetup{text-font-command = \fontfamily{pplx}\selectfont}
  \qty{123456789}{\kilo\volt\per\centi\metre}
\end{LaTeXdemo}

\DescribeOption{text-subscript-command}
\DescribeOption{text-superscript-command}
In most cases, the commands \cs{textsubscript} and \cs{textsuperscript} are
appropriate for creating sub- and superscript material in text mode. However,
when the \pkg{realscripts} package is loaded, it is possible that the output is
not as desired. These two commands are therefore available to allow tuning of
the results. They can also be used for non-standard effects, for example as
here adding color to subscripts.
\begin{LaTeXdemo}
  \sisetup{unit-mode = text}
  \unit{\kg\of{polymer}} \\
  \newcommand*\mysubscript[1]{%
    \textsubscript{\textcolor{blue}{#1}}%
  }
  \sisetup{text-subscript-command = \mysubscript}
  \unit{\kg\of{polymer}}
\end{LaTeXdemo}

\DescribeOption{color}
\DescribeOption{number-color}
\DescribeOption{unit-color}
The color of printed output can be set using the \opt{color} option. When no
color is given, printing follows the surrounding text. In contrast, when a
specific color is given, it is used irrespective of the surroundings. As with
\opt{mode}, the \opt{color} setting may also be applied to numbers and units
independently.
\begin{LaTeXdemo}
  \color{red}%
  Some text \\
  \qty{4}{\kilogram} \\
  More text \\
  \qty[color = blue]{4}{\kilogram} \\
  Still red here!
\end{LaTeXdemo}

\subsection{Parsing numbers}

The package uses a sophisticated parsing system to understand numbers. This
allows \pkg{siunitx} to carry out a range of formatting, as described later.
All of the input options take lists of literal tokens, and are summarised in
Table~\ref{tab:opt:num:in}.
\begin{table}
  \caption{Options for number parsing.%
    \label{tab:opt:num:in}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      evaluate-expression            & Switch  & false            \\
      expression                     & Literal & |#1|             \\^^A (
      input-close-uncertainty        & Literal & )                \\
      input-comparators              & Literal &
        <=>\cs{approx}\cs{ge}\cs{geq} \\
        & & \cs{gg}\cs{le}\cs{leq}\cs{ll}\cs{sim} \\
      input-decimal-markers          & Literal & .,               \\
      input-digits                   & Literal & 0123456789       \\
      input-exponent-markers         & Literal & dDeE             \\
      input-ignore                   & Literal & \meta{none}      \\
      input-open-uncertainty         & Literal & (                \\ ^^A )
      input-signs                    & Literal & +-\cs{pm}\cs{mp} \\
      input-uncertainty-divider      & Literal & :                \\
      input-uncertainty-signs        & Literal & \cs{pm}          \\
      parse-numbers                  & Switch  & true             \\
      retain-explicit-decimal-marker & Switch  & false            \\
      retain-explicit-plus           & Switch  & false            \\
      retain-negative-zero           & Switch  & false            \\
      retain-zero-uncertainty        & Switch  & false            \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{input-digits}
\DescribeOption{input-decimal-markers}
\DescribeOption{input-signs}
\DescribeOption{input-exponent-markers}
The basic parts of a number are the digits, any sign and a separator between
the integer and decimal parts. These are stored in the input options
\opt{input-digits}, \opt{input-decimal-markers} and \opt{input-signs},
respectively. More than one input decimal marker can be used: it will be
converted by the package to the appropriate output marker. Numbers which
include an exponent part also require a marker for the exponent: this again is
taken from the range of tokens in the \opt{input-exponent-markers} option.

\DescribeOption{input-ignore}
Tokens given in the \opt{input-ignore} list are totally passed over by
\pkg{siunitx}: they will be removed from the input with no further processing.

\DescribeOption{input-comparators}
In addition to signs, \pkg{siunitx} can recognise comparators, such as |<|. The
package will automatically carry out conversions for |<<|, |>>|, |<=|, |>=| and
|~| to |\ll|, |\gg|, |\le|, |\ge| and |\sim|, respectively.
\begin{LaTeXdemo}
  \num{< 10} \\
  \qty{>> 5}{\metre} \\
  \num{\le 0.12}
\end{LaTeXdemo}

\DescribeOption{input-open-uncertainty}
\DescribeOption{input-close-uncertainty}
\DescribeOption{input-uncertainty-signs}
In some fields, it is common to give the uncertainty in a number in brackets
after the main part of the number, for example \enquote{\num{1.234(5)}}. The
opening and closing symbols used for this type of input are set as
\opt{input-open-uncertainty} and \opt{input-close-uncertainty}. Alternatively,
the uncertainty may be given as a separate part following a sign. Which signs
are valid for this operation is determined by the \opt{input-uncertainty-signs}
option. As with other signs, the combination |+-| will automatically be
converted to |\pm| internally.
\begin{LaTeXdemo}
  \num{9.99(9)}       \\
  \num{9.99 +- 0.09}  \\
  \num{9.99 \pm 0.09} \\
  \num{123 +- 4.5}    \\
  \num{12.3 +- 6}
\end{LaTeXdemo}

Uncertainties which cross the decimal marker may be given with or without a
decimal marker in \enquote{compact} form. These are treated as equivalent by
the code.\footnote{The package author favors the form without a decimal marker,
and formal guidance is ambiguous on which is correct. The form with a decimal
marker is seen in for example some \acro{NIST} publications.} Where multiple
uncertainty values cross the decimal marker, they should all have the same
number of digits.
\begin{LaTeXdemo}
  \num{123.4(12)} \\
  \num{123.4(1.2)}
\end{LaTeXdemo}
Multiple symmetrical uncertainties may also be given: these must all either be
in the short or the long form.
\begin{LaTeXdemo}
  \num{123.4(12)(45)} \\
  \num{123.4 \pm 1.2 \pm 4.5}
\end{LaTeXdemo}

\DescribeOption{input-uncertainty-divider}
In some areas, uncertainties (or tolerances) are given in an asymmetric format.
This is supported in the \enquote{compact} input form in \pkg{siunitx}, with
the positive and negative parts of the uncertainty divided by a symbol listed
in \opt{input-uncertainty-divider}. When given in this divided form, the
two parts are interpreted as first the positive and then the negative
component. Asymmetric uncertainties which cross the decimal marker are
supported provided both components have the same number of digits.
\begin{LaTeXdemo}
  \num{10.56(12:34)} \\
  \num{123.4(4.6:7.8)}
\end{LaTeXdemo}
Symmetrical and asymmetrical uncertainties may be intermixed.
\begin{LaTeXdemo}
  \num{10.56(1:2)(3)} \\
  \num{6.45(2)(3:4)}
\end{LaTeXdemo}

\DescribeOption{parse-numbers}
The \opt{parse-numbers} option turns the entire parsing system on and off. The
option is made available for two reasons. First, if all of the numbers in a
document are to be reproduced \enquote{as given}, turning off the parser will
represent a significant saving in processing required. Second, it allows the
use of arbitrary \TeX{} code in numbers. If the parser is turned off, the input
will be printed in math mode (requiring |\text| to protect any text in the
number).
\begin{LaTeXdemo}
  \num[parse-numbers = false]{\sqrt{2}}        \\
  \qty[parse-numbers = false]{\sqrt{3}}{\metre}
\end{LaTeXdemo}

\DescribeOption{evaluate-expression}
\DescribeOption{expression}
With the standard settings, numerical input is parsed \enquote{as is} with no
attempt to interpret it mathematically. By enabling the
\opt{evaluate-expression} option, the input can be processed by the standard
\LaTeX3 \acro{FPU} (see package \pkg{xfp} for more). The nature of the
expression itself can be adjusted using the \opt{expression} setting: as
standard, the entire input is simply parsed with no change, but this setting
may be used to add additional steps. The \emph{input} in such an expression is
represented by |#1|. Note that the \acro{FPU} uses its own syntax for numbers,
most notably in that a decimal marker must be |.|.
\begin{LaTeXdemo}
  \sisetup{evaluate-expression}%
  \qty{2 + 4 * 3}{\joule} \\
  \qty[expression = 10 * (#1)]{2 + 4 * 3}{\joule}
\end{LaTeXdemo}

\DescribeOption{retain-explicit-decimal-marker}
\DescribeOption{retain-explicit-plus}
\DescribeOption{retain-negative-zero}
\DescribeOption{retain-zero-uncertainty}
In some areas, a trailing decimal marker with no decimal part present is used
to show that zeros in the integer part are significant. This can be enabled
using the \opt{retain-explicit-decimal-marker} option. The inclusion of a
leading plus sign is usually unnecessary for positive numbers, and so they are
not retained as-standard when parsing. The \opt{retain-explicit-plus} option is
available to control this behaviour. Similarly, an uncertainty of zero is
normally not meaningful, and so is ignored by the parser. This can be
controlled using the \opt{retain-zero-uncertainty} option. Finally, a negative
sign for an entirely zero value may or may not have significance: this is
controlled by the \opt{retain-negative-zero} option.
\begin{LaTeXdemo}
  \num{10.} \\
  \num[retain-explicit-decimal-marker]{10.} \\
  \num{+345} \\
  \num[retain-explicit-plus]{+345} \\
  \num{-0} \\
  \num[retain-negative-zero]{-0} \\
  \num{12.3(0)} \\
  \num[retain-zero-uncertainty]{12.3(0)}
\end{LaTeXdemo}

\subsection{Post-processing numbers}

Before typesetting numbers, various post-processing steps can be carried out.
These involve adding or removing information from the number in a systematic
way; the options are summarised in Table~\ref{tab:opt:num:post}.
\begin{table}
  \caption{Number post-processing options.%
    \label{tab:opt:num:post}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      drop-exponent               & Switch  & false   \\
      drop-uncertainty            & Switch  & false   \\
      drop-zero-decimal           & Switch  & false   \\
      exponent-mode               & Switch  & input   \\
      exponent-thresholds         & Literal & -3:3    \\
      fixed-exponent              & Integer & 0       \\
      minimum-integer-digits      & Integer & 0       \\
      minimum-decimal-digits      & Integer & 0       \\
      round-direction             & Choice  & nearest \\
      round-half                  & Choice  & up      \\
      round-minimum               & Literal & 0       \\
      round-mode                  & Choice  & none    \\
      round-pad                   & Switch  & true    \\
      round-precision             & Integer & 2       \\
      round-zero-positive         & Switch  & true    \\
      uncertainty-round-direction & Choice  & nearest \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{exponent-mode}
\DescribeOption{fixed-exponent}
Numbers can be converted to scientific notation by the package. This is
controlled by the \opt{exponent-mode} option, which takes choices \opt{input},
\opt{fixed}, \opt{engineering}, \opt{scientific} and \opt{threshold}. The
\opt{fixed} setting will use the exponent value by the \opt{fixed-exponent}
option. When \opt{engineering} is set, the exponent is always a power of three.
\begin{LaTeXdemo}
  \num{0.001}  \\
  \num{0.0100} \\
  \num{1200}   \\
  \sisetup{exponent-mode = scientific}%
  \num{0.001}  \\
  \num{0.0100} \\
  \num{1200}   \\
  \sisetup{exponent-mode = engineering}%
  \num{0.001}  \\
  \num{0.0100} \\
  \num{1200}   \\
  \sisetup{
    exponent-mode  = fixed,
    fixed-exponent = 2,
  }%
  \num{0.001}  \\
  \num{0.0100} \\
  \num{1200}
\end{LaTeXdemo}
When used with a \opt{fixed-exponent} of zero, this may be used to remove
scientific notation from the input
\begin{LaTeXdemo}
  \num{1.23e4} \\
  \num[exponent-mode = fixed, fixed-exponent = 0]{1.23e4}
\end{LaTeXdemo}
\DescribeOption{exponent-thresholds}
When the \opt{exponent-mode} is set to \opt{threshold}, values outside of a
threshold range for the exponent are always printed in scientific form. The
threshold range itself is controlled by \opt{exponent-thresholds}, which is
given as \texttt{\meta{min}:\meta{max}} (Table~\ref{tab:threshold}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Thresholds for exponents.%
      \label{tab:threshold}}
    \begin{tabular}
      {
        @{}
        S
        S[exponent-mode = threshold]
        S[exponent-mode = threshold, exponent-thresholds = -2:2]
        @{}
      }
    \toprule
      {Input} & {Threshold $-3:3$} &  {Threshold $-2:2$} \\
    \midrule
      0.001 & 0.001 & 0.001 \\
      0.012 & 0.012 & 0.012 \\
      0.123 & 0.123 & 0.123 \\
      1     & 1     & 1     \\
      12    & 12    & 12    \\
      123   & 123   & 123   \\
      1234  & 1234  & 1234  \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

Exponent mode applies \emph{after} rounding, such that the number of decimal
places for rounding is those which appear in the output.

\DescribeOption{drop-exponent}
\DescribeOption{drop-uncertainty}
The use of an uncertainty can be suppressed entirely using the
\opt{drop-uncertainty} option: this applies \emph{before} rounding is
attempted. Similarly, exponents can be dropped using \opt{drop-exponent} can be
used to suppress the exponent part (\emph{after} conversion to a fixed
exponent).
\begin{LaTeXdemo}
  \num{0.01(2)} \\
  \num[drop-uncertainty]{0.01(2)} \\
   \num{0.01e3} \\
  \num[drop-exponent]{0.01e3}
\end{LaTeXdemo}

\DescribeOption{round-mode}
\DescribeOption{round-precision}
\DescribeOption{round-pad}
The package can round numerical input to a fixed number of significant figures
or decimal places. This is controlled by the \opt{round-mode} option, which
takes the choices \opt{none}, \opt{figures}, \opt{places} and
\opt{uncertainty}. When rounding is turned on, the number of digits used
(either decimal places or significant figures in the mantissa) is set using the
\opt{round-precision} option. Rounding numbers with uncertainties may be
carried out using the \opt{uncertainty} setting to \opt{round-mode}. In this
case the precision is used first to round the uncertainty itself (to a number
of figures), before rounding the main value to follow.
\begin{LaTeXdemo}
  \num{1.23456}     \\
  \num{14.23}      \\
  \num{0.12345(9)} \\
  \sisetup{
    round-mode      = places,
    round-precision = 3
  }%
  \num{1.23456}     \\
  \num{14.23}      \\
  \num{0.12345(9)} \\
  \sisetup{
    round-mode      = figures,
    round-precision = 3
  }%
  \num{1.23456}     \\
  \num{14.23}      \\
  \num{0.12345(9)} \\
  \sisetup{
    round-mode      = uncertainty,
    round-precision = 1
  }%
  \num{0.12345(9)}   \\
  \num{0.12345(23)}  \\
  \num{0.12345(234)}
\end{LaTeXdemo}
Rounding may \enquote{extend} a short number to more digits (or figures): this
is controlled by the switch \opt{round-pad}, which is \opt{true} as standard.
\begin{LaTeXdemo}
  \sisetup{round-mode = figures, round-precision = 4}%
  \num{12.3} \\
  \num[round-pad = false]{12.3}
\end{LaTeXdemo}

\DescribeOption{round-direction}
\DescribeOption{round-half}
The detail of which approach to use for rounding may be adjusted using the
options \opt{round-direction} and \opt{round-half}. The option
\opt{round-direction} determines which direction a value is rounded toward: a
choice of \opt{nearest}, \opt{up} or \opt{down}. The standard setting,
\opt{nearest}, gives the common outcome that values round depending on whether
the preceding digit is greater or less than \num{5}. The options \opt{up} and
\opt{down} mean that values are always rounded in one direction: the \opt{down}
option may be thought of as \enquote{truncation}.
\begin{LaTeXdemo}
  \sisetup{round-mode = places}
  \num{0.054} \\
  \num{0.046} \\
  \sisetup{round-direction = down}%
  \num{0.054} \\
  \num{0.046} \\
  \sisetup{round-direction = up}%
  \num{0.054} \\
  \num{0.046}
\end{LaTeXdemo}
For rounding to \opt{nearest}, where the rounded part of a number is exactly
half, there are two common methods for \enquote{breaking the tie}. The choice
of method is determined by the option \opt{round-half}, which recognises the
choices \opt{up} and \opt{even}.
\begin{LaTeXdemo}
  \sisetup{
    round-mode      = figures,
    round-precision = 1,
    round-half      = up
  }%
  \num{0.055} \\
  \num{0.045} \\
  \sisetup{round-half = even}%
  \num{0.055} \\
  \num{0.045}
\end{LaTeXdemo}

\DescribeOption{uncertainty-round-direction}
When rounding with uncertainties, the direction to round the uncertainty part
may be set independently of the main number: this is typically used to round
the uncertainty strictly upward.
\begin{LaTeXdemo}
  \sisetup{round-mode = uncertainty}
  \num{0.123(41)} \\
  \sisetup{uncertainty-round-direction = up}%
  \num{0.123(41)}
\end{LaTeXdemo}

\DescribeOption{round-minimum}
There are cases in which rounding will result in the number reaching zero. It
may be desirable to show such results as below a threshold value. This can be
achieved by setting \opt{round-minimum} to the threshold value. There will be
no effect when rounding to a number of significant figures as it is not
possible to obtain the value zero in these cases.
\begin{LaTeXdemo}
  \sisetup{round-mode = places}%
  \num{0.0055} \\
  \num{0.0045} \\
  \sisetup{round-minimum = 0.01}%
  \num{0.0055} \\
  \num{0.0045}
\end{LaTeXdemo}

\DescribeOption{round-zero-positive}
When rounding negative numbers to a fixed number of places, a zero value may
result. Usually this is expressed as an unsigned value, but in some cases
retaining the negative sign may be desirable. This behaviour can be controlled
using the \opt{round-zero-positive} switch.
\begin{LaTeXdemo}
  \sisetup{round-mode = places}%
  \num{-0.001} \\
  \sisetup{round-zero-positive = false}%
  \num{-0.001}
\end{LaTeXdemo}

\DescribeOption{drop-zero-decimal}
It may be desirable to convert decimals to integers when the decimal part is
zero. This is set up using the \opt{drop-zero-decimal} option, which applies
after rounding but before setting minimum numbers of digits.
\begin{LaTeXdemo}
  \num{2.0} \\
  \num{2.1} \\
  \sisetup{drop-zero-decimal}%
  \num{2.0} \\
  \num{2.1}
\end{LaTeXdemo}

\DescribeOption{minimum-decimal-digits}
\DescribeOption{minimum-integer-digits}
The \opt{minimum-decimal-digits} and \opt{minimum-integer-digits} option may be
used to pad numbers to a given size. This applies independent of any rounding.
\begin{LaTeXdemo}
  \num{123} \\
  \num[minimum-integer-digits = 2]{123} \\
  \num[minimum-integer-digits = 4]{123} \\
  \num{0.123} \\
  \num[minimum-decimal-digits = 2]{0.123} \\
  \num[minimum-decimal-digits = 4]{0.123} \\
\end{LaTeXdemo}

\subsection{Printing numbers}

Actually printing numbers is controlled by a number of settings, which apply
ideas such as differing decimal markers, digit grouping and so on. All of these
options are concerned with the appearance of output, rather than the data it
conveys. The options are summarised in Table~\ref{tab:opt:num:out}.
\begin{table}
  \caption{Output options for numbers.%
    \label{tab:opt:num:out}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{l}{Option name} &
      Type &
      \multicolumn{1}{l}{Default} \\
    \midrule
      bracket-negative-numbers         & Switch  & false             \\
      digit-group-size                 & Integer & 3                 \\
      digit-group-first-size           & Integer & 3                 \\
      digit-group-other-size           & Integer & 3                 \\
      exponent-base                    & Literal & 10                \\
      exponent-product                 & Math    & \verb=\times=     \\
      group-digits                     & Choice  & all               \\
      group-minimum-digits             & Integer & 5                 \\
      group-separator                  & Literal & \cs{,}            \\
      negative-color                   & Literal & \meta{none}       \\ ^^A (
      output-close-uncertainty         & Literal & )                 \\
      output-decimal-marker            & Literal & .                 \\
      output-exponent-marker           & Literal & \meta{none}       \\
      output-open-uncertainty          & Literal & (                 \\ ^^A )
      print-exponent-implicit-plus     & Switch  & false             \\
      print-implicit-plus              & Switch  & false             \\
      print-mantissa-implicit-plus     & Switch  & false             \\
      print-unity-mantissa             & Switch  & true              \\
      print-zero-exponent              & Switch  & false             \\
      print-zero-integer               & Switch  & true              \\
      simplify-uncertainty             & Switch  & false             \\
      tight-spacing                    & Switch  & false             \\
      uncertainty-descriptor-mode      & Choice  & bracket-separator \\
      uncertainty-descriptor-separator & Literal & \cs{ }            \\
      uncertainty-descriptors          & Literal & \meta{none}       \\
      uncertainty-mode                 & Choice  & compact           \\
      uncertainty-separator            & Literal & \meta{none}       \\
      zero-decimal-as-symbol           & Switch  & false             \\
      zero-symbol                      & Literal & \verb=\mbox{---}= \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{group-digits}
\DescribeOption{group-separator}
Grouping digits into blocks of three is a common method to increase the ease of
reading of numbers. The \opt{group-digits} choice controls whether this
behaviour applies, and takes the values \opt{all}, \opt{none}, \opt{decimal}
and \opt{integer}. Grouping can be activated separately for the integer and
decimal parts of a number using the appropriately-named values.
\begin{LaTeXdemo}
  \num{12345.67890} \\
  \num[group-digits = none]{12345.67890}   \\
  \num[group-digits = decimal]{12345.67890} \\
  \num[group-digits = integer]{12345.67890}
\end{LaTeXdemo}
The separator used between groups of digits is stored by the
\opt{group-separator} option. This takes literal input and may be used in math
mode: for a text-mode full space use \verb*|\ |.
\begin{LaTeXdemo}
  \num{12345} \\
  \num[group-separator = {,}]{12345} \\
  \num[group-separator = \ ]{12345}
\end{LaTeXdemo}

\DescribeOption{group-minimum-digits}
Grouping is not always applied to smaller numbers; this can be controlled using
the option \opt{group-minimum-digits}, which specifies how many digits must be
present before grouping is applied. The number of digits is considered
separately for the integer and decimal parts of the number: grouping does not
\enquote{cross the boundary}.
\begin{LaTeXdemo}
  \num{1234} \\
  \num{12345} \\
  \num[group-minimum-digits = 5]{1234} \\
  \num[group-minimum-digits = 5]{12345} \\
  \num{1234.5678} \\
  \num{12345.67890} \\
  \num[group-minimum-digits = 5]{1234.5678} \\
  \num[group-minimum-digits = 5]{12345.67890}
\end{LaTeXdemo}

\DescribeOption{digit-group-size}
\DescribeOption{digit-group-first-size}
\DescribeOption{digit-group-other-size}
The number of digits in each group can be controlled by the setting
\opt{digit-group-size}, which has standard value $3$. Finer control can be
achieved using \opt{digit-group-first-size} and \opt{digit-group-other-size}:
the first group is that immediately by the decimal point, the other value
applies to the second and subsequent groupings. These can be used for
example to achieve the grouping typically used in India (lakh).
\begin{LaTeXdemo}
  \num{1234567890} \\
  \num[digit-group-size = 5]{1234567890} \\
  \num[digit-group-other-size = 2]{1234567890}
\end{LaTeXdemo}

\DescribeOption{output-decimal-marker}
The decimal marker used in output is set using the \opt{output-decimal-marker}
option; this can differ from the input marker.
\begin{LaTeXdemo}
  \num{1.23} \\
  \num[output-decimal-marker = {,}]{1.23} \\
\end{LaTeXdemo}

\DescribeOption{exponent-base}
\DescribeOption{exponent-product}
When exponents are present in the input, the \opt{exponent-base} and
\opt{exponent-product} options set the obvious parts of the output.
\begin{LaTeXdemo}
  \num[exponent-product = \times]{1e2} \\
  \num[exponent-product = \cdot]{1e2} \\
  \num[exponent-base = 2]{1e2}
\end{LaTeXdemo}
\DescribeOption{output-exponent-marker}
Alternatively, if the \opt{output-exponent-marker} option is set then the value
stored will be used in place of the normal product and base combination.
\begin{LaTeXdemo}
  \num[output-exponent-marker = e]{1e2} \\
  \num[output-exponent-marker = \mathrm{E}]{1e2}
\end{LaTeXdemo}

\DescribeOption{uncertainty-mode}
\DescribeOption{output-open-uncertainty}
\DescribeOption{output-close-uncertainty}
\DescribeOption{uncertainty-separator}
When input is given including a single uncertainty, it can be printed either
with the uncertainty in brackets or as a separate number. This behaviour is
controlled by the \opt{uncertainty-mode} choice. When this is set to
\opt{separate}, the uncertainty is printed as an entirely separate number
preceded by \cs{pm}. Other settings all place the uncertainty in brackets
directly attached to the main value. The standard setting of \opt{compact}
prints digits of uncertainty in the least-significant digits. It does
\emph{not} print a decimal marker if the uncertainty crosses the decimal. The
setting \opt{full} prints the full value of the uncertainty. Finally,
\opt{compact-marker} is available to print in the \opt{compact} style except
where the uncertainty crosses the decimal, in which case the \opt{full} style
is used. When the uncertainty is given in brackets, a space may be added
between the main number and the uncertainty: this is stored using the
\opt{uncertainty-separator} option. The opening and closing brackets used are
stored in \opt{output-open-uncertainty} and \opt{output-close-uncertainty},
respectively. Tokens may be inserted before the opening bracket using
\opt{uncertainty-separator}.
\begin{LaTeXdemo}
  \num{123.45(120)} \\
  \num{0.035(14)}   \\
  \sisetup{uncertainty-mode = full}
  \num{123.45(120)} \\
  \num{0.035(14)}   \\
  \sisetup{uncertainty-mode = compact-marker}
  \num{123.45(120)} \\
  \num{0.035(14)}   \\
  \sisetup{uncertainty-mode = separate}
  \sisetup{
    output-open-uncertainty  = [,
    output-close-uncertainty = ],
    uncertainty-separator    = \,
  }%
  \num{1.234(5)}
\end{LaTeXdemo}

\DescribeOption{uncertainty-descriptors}
\DescribeOption{uncertainty-descriptor-mode}
\DescribeOption{uncertainty-descriptor-separator}
Multiple uncertainties can be given for a number. These are always printed in a
separated form. When there is more than one uncertainty part, it may be useful
to describe the nature of this value. This can be achieved using the
\opt{uncertainty-descriptors} option, which take a comma-separated list of
descriptions. The formatting of the descriptors can be adjusted using the
settings \opt{uncertainty-descriptor-mode} and
\opt{uncertainty-descriptor-separator}. The choices for the \opt{mode} are
\opt{bracket}, \opt{bracket-separator}, \opt{separator} and \opt{subscript}.
\begin{LaTeXdemo}
  \num{1.2(3)(4)} \\
  \sisetup{uncertainty-descriptors = {sys, stat}}
  \num{1.2(3)(4)} \\
  \num[uncertainty-descriptor-mode = subscript]{1.2(3)(4)}
\end{LaTeXdemo}

\DescribeOption{simplify-uncertainty}
Where the upper and lower parts of an asymmetrical uncertainty are identical,
it may be desirable to print as in symmetrical format. This can be enabled
using the \opt{simplify-uncertainty} option.
\begin{LaTeXdemo}
  \num{10.56(2:2)} \\
  \num[simplify-uncertainty]{10.56(2:2)}
\end{LaTeXdemo}

\DescribeOption{bracket-ambiguous-numbers}
There are certain combinations of numerical input which can be ambiguous. This
can be corrected by adding brackets in the appropriate place, and is controlled
by the \opt{bracket-ambiguous-numbers} switch. This option only applies to pure
numbers: when formatting quantities, the need for brackets also depends on the
placement of units, so is controlled by \opt{separate-uncertainty-units}.
\begin{LaTeXdemo}
  \sisetup{uncertainty-mode = separate}
  \num{1.2(3)e4} \\
  \num[bracket-ambiguous-numbers = false]{1.2(3)e4}
\end{LaTeXdemo}

\DescribeOption{negative-color}
The package can detect negative mantissa values and alter print color
accordingly. This is disabled by setting the option to an empty value.
\begin{LaTeXdemo}
  \num{-15673} \\
  \num[negative-color = red]{-15673}
\end{LaTeXdemo}

\DescribeOption{bracket-negative-numbers}
A common means to display negative numbers in financial situations is to place
them in brackets. This can be carried out automatically using the
\opt{bracket-negative-numbers} option.
\begin{LaTeXdemo}
  \num{-15673} \\
  \num[bracket-negative-numbers]{-15673} \\
  \qty{-10}{\metre} \\
  \qty[bracket-negative-numbers]{-10}{\metre}
\end{LaTeXdemo}

\DescribeOption{tight-spacing}
Under some circumstances is may be desirable to \enquote{squeeze} the output
spacing. This is turned on using the \opt{tight-spacing} switch, which
compresses spacing where possible.
\begin{LaTeXdemo}
  \num{2e3} \\
  \num[tight-spacing = true]{2e3}
\end{LaTeXdemo}

\DescribeOption{print-implicit-plus}
\DescribeOption{print-mantissa-implicit-plus}
\DescribeOption{print-exponent-implicit-plus}
It may be useful to force all numbers to have a sign. This behaviour is
controlled by the \opt{print-implicit-plus} option: this is used if given and
if no sign was present in the input. It is possible to set this behavior for
the exponent and mantissa independently.
\begin{LaTeXdemo}
  \num{345} \\
  \num[print-implicit-plus]{345}
\end{LaTeXdemo}

\DescribeOption{print-unity-mantissa}
\DescribeOption{print-zero-exponent}
\DescribeOption{print-zero-integer}
Printing of a mantissa of $1$, an exponent of $0$ and an integer component of
$0$ is controllable by the options \opt{print-unity-mantissa},
\opt{print-zero-exponent} and \opt{print-zero-integer}. The standard settings
print a mantissa of $1$ and an integer part of $0$, but omit an exponent of
$0$. Notice that where both \opt{print-unity-mantissa} and
\opt{print-zero-exponent} are set to \opt{false}, the value $1$ will still be
printed (\foreign{i.e.}~\opt{print-zero-exponent} has a higher priority).
\begin{LaTeXdemo}
  \num{1e4} \\
  \num[print-unity-mantissa = false]{1e4} \\
  \num{444e0} \\
  \num[print-zero-exponent = true]{444e0} \\
  \num{0.123} \\
  \num[print-zero-integer = false]{0.123}
\end{LaTeXdemo}

\DescribeOption{zero-decimal-as-symbol}
\DescribeOption{zero-symbol}
In some areas, particularly financial, entirely zero decimal parts are
replaced by a dash. This is supported by option \opt{zero-decimal-as-symbol},
which then uses the material stored using \opt{zero-symbol} as the replacement.
\begin{LaTeXdemo}
  \num{123.00} \\
  \sisetup{zero-decimal-as-symbol}
  \num{123.00} \\
  \num[zero-symbol = \text{[{---}]}]{123.00}
\end{LaTeXdemo}

\subsection{Lists, products and ranges}

Lists, products and ranges of numbers and quantities have a small number of
specialised options, which apply to these more unusual input forms
(Table~\ref{tab:opt:num:list}).
\begin{table}
  \begin{threeparttable}
    \caption{Output options for lists, products and ranges of numbers and
      quantities.%
      \label{tab:opt:num:list}}
    \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
      \toprule
        \multicolumn{1}{l}{Option name} &
        Type &
        \multicolumn{1}{l}{Default\tnote{*}} \\
      \midrule % (
        list-close-bracket         & Literal & )                    \\
        list-exponents             & Choice  & individual           \\
        list-final-separator       & Literal & \verb*= \text{and} = \\
        list-independent-prefix    & Switch  & false                \\
        list-open-bracket          & Literal & (                    \\ % )
        list-pair-separator        & Literal & \verb*= \text{and} = \\
        list-separator             & Literal & \verb*=\text{,} =    \\
        list-units                 & Choice  & repeat               \\ % (
        product-close-bracket      & Literal & )                    \\
        product-exponents          & Choice  & individual           \\
        product-independent-prefix & Switch  & false                \\
        product-mode               & Choice  & symbol               \\
        product-open-bracket       & Literal & (                    \\ % )
        product-phrase             & Literal & \verb*= \text{by} =  \\
        product-symbol             & Literal & \cs{times}           \\
        product-units              & Choice  & repeat               \\ % (
        range-close-bracket        & Literal & )                    \\
        range-exponents            & Choice  & individual           \\
        range-independent-prefix   & Switch  & false                \\
        range-open-bracket         & Literal & (                    \\ % )
        range-open-phrase          & Literal & \meta{empty}         \\
        range-phrase               & Literal & \verb*= \text{to} =  \\
        range-units                & Choice  & repeat               \\
      \bottomrule
    \end{tabular}
    \footnotesize
    \begin{tablenotes}
      \item[*] The default values are actually more complex for two
        reasons: allowing spaces to work in both math and text modes,
        and localization of strings.
    \end{tablenotes}
  \end{threeparttable}
\end{table}

\DescribeOption{list-final-separator}
\DescribeOption{list-pair-separator}
\DescribeOption{list-separator}
Lists of numbers are printed with a separator between each item, which is
stored using the \opt{list-separator} option. The separator before the last
item of a list may be different, and is therefore set using the
\opt{list-final-separator} option. The separator used for exactly two items is
set using the \opt{list-pair-separator} option. Any spaces needed should be
included in the option settings: none are added within the code. The separators
are always printed in text mode.
\begin{LaTeXdemo}
  \numlist{0.1;0.2;0.3}                              \\
  \numlist[list-separator = {; }]{0.1;0.2;0.3}       \\
  \numlist[list-final-separator = {, }]{0.1;0.2;0.3} \\
  \numlist[
    list-separator       = { and },
    list-final-separator = { and finally }
  ]{0.1;0.2;0.3} \\
  \numlist{0.1;0.2} \\
  \numlist[list-pair-separator = {, and }]{0.1;0.2}
\end{LaTeXdemo}

\DescribeOption{product-mode}
\DescribeOption{product-phrase}
\DescribeOption{product-symbol}
Products of numbers can be output using either a product symbol or phrase:
this is controlled by the \opt{product-mode} setting. When \opt{symbol} is
set, the appropriate symbol is stored in \opt{product-symbol}. When using
\opt{phrase-mode}, the information is stored in \opt{product-phrase}. Phrases
are always printed in text mode; symbols are printed using the same routine
as for numbers.
\begin{LaTeXdemo}
  \numproduct{5 x 100 x 2} \\
  \numproduct[product-symbol = \ensuremath{\cdot}]{5 x 100 x 2} \\
  \sisetup{product-mode = phrase}%
  \numproduct{5 x 100 x 2}\\
  \numproduct[product-phrase = { BY }]{5 x 100 x 2} \\
\end{LaTeXdemo}

\DescribeOption{range-open-phrase}
\DescribeOption{range-phrase}
Ranges of numbers can be given as input. These will have an appropriate word or
symbol inserted between the two entries: this is stored using the
\opt{range-phrase} option. The phrase should include any necessary spaces: no
extra space is added. The phrase is always printed in text mode.
\begin{LaTeXdemo}
  \numrange{5}{100} \\
  \numrange[range-phrase = --]{5}{100}
\end{LaTeXdemo}
In some languages, an opening phrase is \emph{required}; this is supported
using the \opt{range-open-phrase} key.
\begin{LaTeXdemo}
  \numrange{10}{12} \\
  \numrange[range-open-phrase = {\text{from} }]{5}{100}
\end{LaTeXdemo}

\DescribeOption{list-exponents}
\DescribeOption{product-exponents}
\DescribeOption{range-exponents}
Lists, products and ranges can be \enquote{compressed} by combining the
exponent parts. This is controlled by the options \opt{list-exponents},
\opt{product-exponents} and \opt{range-exponents}, all of which take the
choices \opt{individual}, \opt{combine-bracket} and \opt{combine}. The standard
setting, \opt{individual}, leaves the exponent with the matching value. Both
\opt{combine} and \opt{combine-bracket} take the exponent of the first entry
and apply to to all other entries, with the exponent itself places at the end.
\begin{LaTeXdemo}
  \numlist{5e3;7e3;9e3;1e4} \\
  \numproduct{5e3 x 7e3 x 9e3 x 1e4} \\
  \numrange{5e3}{7e3} \\
  \sisetup
    {
      list-exponents    = combine-bracket ,
      product-exponents = combine-bracket ,
      range-exponents   = combine-bracket
    }
  \numlist{5e3;7e3;9e3;1e4} \\
  \numproduct{5e3 x 7e3 x 9e3 x 1e4} \\
  \numrange{5e3}{7e3} \\
  \sisetup
    {
      list-exponents    = combine ,
      product-exponents = combine ,
      range-exponents   = combine
    }
  \numlist{5e3;7e3;9e3;1e4} \\
  \numproduct{5e3 x 7e3 x 9e3 x 1e4} \\
  \numrange{5e3}{7e3}
\end{LaTeXdemo}

\DescribeOption{list-units}
\DescribeOption{product-units}
\DescribeOption{range-units}
The \opt{list-units}, \opt{product-units} and \opt{range-units} options
determine how \cs{qtylist}, \cs{qtyproduct} and \cs{qtyrange} command print
units, respectively. The standard setting for these is \opt{repeat}, where each
number will be printed with a unit. Alternatives are \opt{bracket} and
\opt{single}. If set to \opt{single}, this will override collection of
exponents.
\begin{LaTeXdemo}
  \qtylist{2;4;6;8}{\tesla} \\
  \qtylist[list-units = bracket]{2;4;6;8}{\tesla} \\
  \qtylist[list-units = repeat]{2;4;6;8}{\tesla} \\
  \qtylist[list-units = single]{2;4;6;8}{\tesla} \\
  \qtyrange{2}{4}{\degreeCelsius} \\
  \qtyrange[range-units = bracket]{2}{4}{\degreeCelsius} \\
  \qtyrange[range-units = repeat]{2}{4}{\degreeCelsius} \\
  \qtyrange[range-units = single]{2}{4}{\degreeCelsius}
\end{LaTeXdemo}
The option \opt{product-units} also offers the settings \opt{bracket-power} and
\opt{power}.
\begin{LaTeXdemo}
  \qtyproduct{2 x 4}{\metre} \\
  \qtyproduct[product-units = bracket-power]{2 x 4}{\metre} \\
  \qtyproduct[product-units = power]{2 x 4}{\metre}
\end{LaTeXdemo}
\DescribeOption{list-close-bracket}
\DescribeOption{list-open-bracket}
\DescribeOption{product-close-bracket}
\DescribeOption{product-open-bracket}
\DescribeOption{range-close-bracket}
\DescribeOption{range-open-bracket}
The brackets used can be customised using the relevant \texttt{\dots-close-bracket}
and \texttt{\dots-open-bracket} options.
\begin{LaTeXdemo}
  \sisetup{
    list-units         = bracket ,
    list-open-bracket  = ( ,
    list-close-bracket = )
  }
  \numlist{5e3;7e3;9e3;1e4} \\
  \sisetup{
    product-units         = bracket ,
    product-open-bracket  = ( ,
    product-close-bracket = )
  }
  \numproduct{5e3 x 7e3 x 9e3 x 1e4} \\
  \sisetup{
    range-units         = bracket ,
    range-open-bracket  = ( ,
    range-close-bracket = )
  }
  \qtyrange{2}{4}{\degreeCelsius}
\end{LaTeXdemo}

\DescribeOption{list-independent-prefix}
\DescribeOption{product-independent-prefix}
\DescribeOption{range-independent-prefix}
When converting exponents to prefixes, there are two possible approaches to the
case where units are repeated. They can all be converted to use the same
(initial) prefix, or they can be processed separately. The
\texttt{\dots-independent-prefix} options are used to control this outcome.
\begin{LaTeXdemo}
  \sisetup{exponent-to-prefix = true}%
  \qtyrange{1e3}{1e9}{\watt}\\
  \sisetup{range-independent-prefix = true}%
  \qtyrange{1e3}{1e9}{\watt}
\end{LaTeXdemo}

\subsection{Complex numbers}

A small number of options apply specifically to the handling of complex
numbers; these are summarised in Table~\ref{tab:opt:num:complex}.
\begin{table}
  \caption{Options for complex numbers.%
    \label{tab:opt:num:complex}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      complex-angle-unit    & Choice  & degrees           \\
      complex-mode          & Choice  & input             \\
      complex-root-position & Choice  & after-number      \\
      complex-symbol-angle  & Literal & \cs{angle}        \\
      complex-symbol-degree & Literal & \cs{degree}       \\
      input-complex-root    & Literal & ij                \\
      output-complex-root   & Literal & \verb=\mathrm{i}= \\
      print-complex-unity   & Switch  & false             \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{complex-mode}
The format in which complex values are printed can be set using the
\opt{complex-mode} option. With the standard setting (\opt{input}), the complex
value is printed as-given. By setting the option to \opt{cartesian} or
\opt{polar}, the output format can be set to an Cartesian or polar form.
Conversion uses the \LaTeX3 floating-point unit, so is limited to $16$ decimal
places. When converting from Cartesian to polar form, the complex root symbol
must come at the \emph{end} of the imaginary part. It must also be specified
using \texttt{i}.
\begin{LaTeXdemo}
  \complexnum{1 + i} \\
  \complexnum{1:45}  \\
  \complexnum[complex-mode = cartesian]{1 + i} \\
  \complexnum[complex-mode = cartesian, round-mode = places]{1:45}  \\
  \complexnum[complex-mode = polar]{1 + i} \\
  \complexnum[complex-mode = polar]{1:45}
\end{LaTeXdemo}

\DescribeOption{input-complex-root}
When using complex numbers in input, the complex root $(\mathrm{i} =
\sqrt{-1}\,)$ is indicated by one of the tokens stored in
\opt{input-complex-roots}. The parser understands complex root symbols given
either before or after the associated number (but will detect any invalid
arrangement):
\begin{LaTeXdemo}
  \complexnum{9.99 + 88.8i} \\
  \complexnum{9.99 + i88.8}
\end{LaTeXdemo}

\DescribeOption{output-complex-root}
The output complex root symbol is independent of the input and can be changed
using the \opt{output-complex-root} setting.
\begin{LaTeXdemo}
  \complexnum[output-complex-root = i]{1+2i} \\
  \complexnum[output-complex-root = j]{1+2i}
\end{LaTeXdemo}

\DescribeOption{complex-root-position}
 The position of the complex root can be adjusted to place it
 either before or after the associated numeral in a complex number
 using the \opt{complex-root-position} option.
\begin{LaTeXdemo}
  \complexnum{67-0.9i} \\
  \complexnum[complex-root-position = before-number]{67-0.9i} \\
  \complexnum[complex-root-position = after-number]{67-0.9i}
\end{LaTeXdemo}

\DescribeOption{complex-angle-unit}
\DescribeOption{complex-symbol-angle}
\DescribeOption{complex-symbol-degree}
When printing or converting to polar form, the angle may be interpreted in
units set by \opt{complex-angle-unit}: one of \opt{degrees} or \opt{radians}.
The symbol used to denote the angle, and that used for units of degrees,
are controlled by the options \opt{complex-symbol-angle} and
\opt{complex-symbol-degree}, respectively.
\begin{LaTeXdemo}
  \complexqty{1:1}{\ohm} \\
  \complexqty[complex-angle-unit = radians]{1:1}{\ohm} \\
  \complexqty[complex-symbol-angle = \mathrm{A}]{1:1}{\ohm} \\
  \complexqty[complex-symbol-degree = d]{1:1}{\ohm}
\end{LaTeXdemo}

\DescribeOption{print-complex-unity}
When the complex part of a number is exactly \num{1}, it is possible to either
print or suppress the value. This is controlled by the switch
\opt{print-complex-unity}.
\begin{LaTeXdemo}
  \complexqty{1i}{\ohm} \\
  \complexqty[print-complex-unity]{1i}{\ohm}
\end{LaTeXdemo}

\subsection{Angles}

Angle processing provided by the \cs{ang} function has a set of options which
apply in addition to the general ones set up for number processing.
\begin{table}
  \caption{Angle options.%
    \label{tab:opt:ang}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      angle-mode                & Choice  & input          \\
      angle-symbol-degree       & Literal & \cs{degree}    \\
      angle-symbol-minute       & Literal & \cs{arcminute} \\
      angle-symbol-over-decimal & Switch  & false          \\
      angle-symbol-second       & Literal & \cs{arcsecond} \\
      angle-separator           & Literal & \meta{empty}   \\
      fill-angle-degrees        & Switch  & false          \\
      fill-angle-minutes        & Switch  & false          \\
      fill-angle-seconds        & Switch  & false          \\
      number-angle-product      & Literal & \meta{empty}   \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{angle-mode}
The format in which angles are printed can be set using the \opt{angle-mode}
option. With the standard setting (\opt{input}), the angle is printed as-given.
By setting the option to \opt{arc} or \opt{decimal}, the output format can
be set to an arc (degrees/minutes/seconds) or decimal value. Conversion uses
the \LaTeX3 floating-point unit, so is limited to $16$ decimal places.
\begin{LaTeXdemo}
  \ang{2.67} \\
  \ang{2;3;4} \\
  \ang[angle-mode = arc]{2.67} \\
  \ang[angle-mode = arc]{2;3;4} \\
  \ang[angle-mode = decimal]{2.67} \\
  \ang[angle-mode = decimal]{2;3;4} \\
\end{LaTeXdemo}
When integer angles are converted to arc format, the presence of minute and
second components is controlled by the options \opt{fill-angle-minutes} and
\opt{fill-angle-seconds} (see below)

\DescribeOption{number-angle-product}
The separator between the number and angle symbol (degrees, minutes or seconds)
can be set using the \opt{number-angle-product} option, independent of the
related \opt{quantity-product} option used by the \cs{qty} command.
\begin{LaTeXdemo}
  \ang{2.67} \\
  \ang[number-angle-product = \,]{2.67}
\end{LaTeXdemo}

\DescribeOption{angle-separator}
When angles are printed in arc format, the separation of the different parts is
set up using the \opt{arc-separator} option.
\begin{LaTeXdemo}
  \ang{6;7;6.5} \\
  \ang[angle-separator = \,]{6;7;6.5}
\end{LaTeXdemo}

\DescribeOption{fill-angle-degrees}
\DescribeOption{fill-angle-minutes}
\DescribeOption{fill-angle-seconds}
Zero-filling for the degree, minute or second parts of an arc is controlled
using the \opt{fill-angle-degrees}, \opt{fill-angle-minutes} and
\opt{fill-angle-seconds} options. All are off as standard.
\begin{LaTeXdemo}
  \ang{-1;;} \\
  \ang{;-2;} \\
  \ang{;;-3} \\
  {
    \sisetup{fill-angle-degrees}
    \ang{-1;;} \\
    \ang{;-2;} \\
    \ang{;;-3} \\
  }
  {
    \sisetup{fill-angle-minutes}
    \ang{-1;;} \\
    \ang{;-2;} \\
    \ang{;;-3} \\
  }
  {
    \sisetup{fill-angle-seconds}
    \ang{-1;;} \\
    \ang{;-2;} \\
    \ang{;;-3}
  }
\end{LaTeXdemo}

\DescribeOption{angle-symbol-degree}
\DescribeOption{angle-symbol-minute}
\DescribeOption{angle-symbol-second}
With the standard settings, the symbols used for arc angles are the unit
commands \cs{degree}, \cs{arcminute} and \cs{arcsecond}. These can be altered
using \opt{angle-symbol-degree}, \opt{angle-symbol-minute} and
\opt{angle-symbol-second}. This is most likely to be used when the definition
of the unit macros is altered, for example to set \cs{arcsecond} as
\texttt{as}.
\begin{LaTeXdemo}
  \ang{6;7;6.5} \\
  \sisetup{
    angle-symbol-degree = d ,
    angle-symbol-minute = m ,
    angle-symbol-second = s
  }
  \ang{6;7;6.5}
\end{LaTeXdemo}

\DescribeOption{angle-symbol-over-decimal}
In some subject areas, most notably astronomy, the angle symbols are given over
the decimal marker, rather than at the end of the number. This behaviour is
available using the \opt{angle-symbol-over-decimal} option.
\begin{LaTeXdemo}
  \ang{45.697}  \\
  \ang{6;7;6.5} \\
  \ang[angle-symbol-over-decimal]{45.697} \\
  \ang[angle-symbol-over-decimal]{6;7;6.5}
\end{LaTeXdemo}

\subsection{Creating units%
  \label{sec:units:creating}}

The various macro units are created at the start of the document. \pkg{siunitx}
can define these such that they are only available for use within the \cs{unit}
and \cs{qty} functions, or can make the unit macros available throughout the
document body. There are a number of settings which control this creation
process (Table~\ref{tab:opt:units:def}). As a result, these options all apply
in the preamble only.
\begin{table}
  \caption{Unit creation options.%
    \label{tab:opt:units:def}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      free-standing-units    & Switch & false \\
      overwrite-command      & Switch & false \\
      space-before-unit      & Switch & false \\
      unit-optional-argument & Switch & false \\
      use-xspace             & Switch & false \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{free-standing-units}
\DescribeOption{overwrite-functions}
The \opt{free-standing-units} option controls whether the unit macros exist
outside of the \cs{unit} and \cs{qty} arguments. When this option is
\opt{true}, \pkg{siunitx} creates the macros for general use. The standard
method to achieve this does not overwrite any existing macros: this behaviour
can be altered using the \opt{overwrite-commands} switch.

When using the approach of \enquote{free-standing} unit commands, only macros
created using \cs{DeclareSIUnit} are defined generally. Thus prefixes and
powers should be combined with the desired unit into a single free-standing
command, for example
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\kilometre{\kilo\metre}
\end{LaTeXdemo}

\DescribeOption{space-before-unit}
\DescribeOption{unit-optional-argument}
\DescribeOption{use-xspace}
When \enquote{free standing} unit macros are created, their behaviour can be
adjusted by a number of options. These are mainly intended for emulating the
input syntax of older packages. The option \opt{unit-optional-argument} gives
the same behaviour for the inputs
\begin{LaTeXdemo}[code only]
  \qty{10}{\metre}
\end{LaTeXdemo}
and
\begin{LaTeXdemo}[code only]
  \metre[10].
\end{LaTeXdemo}
The \opt{space-before-unit} and \opt{use-xspace} options control the behaviour
at the \enquote{ends} of the unit macros. Activating \opt{space-before-unit}
inserts the number--unit space before the unit is printed. This is suitable for
the input syntax
\begin{LaTeXdemo}[code only]
  30\metre
\end{LaTeXdemo}
but does mean that the unit macros are incorrectly spaced in running text. On
the other hand, the \opt{use-xspace} option attempts to correctly space input
such as
\begin{LaTeXdemo}[code only]
  \metre is the symbol for metres.
\end{LaTeXdemo}

\subsection{Using units}

Part of the power of \pkg{siunitx} is the ability to alter the output format
for units without changing the input. The behaviour of units is therefore
controlled by a number of options which alter either the processing of units or
the output directly (Table~\ref{tab:opt:units:out}).
\begin{table}
  \centering
  \caption{Unit output options.%
    \label{tab:opt:units:out}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      bracket-unit-denominator     & Switch  & true         \\
      forbid-literal-units         & Switch  & false        \\
      fraction-command             & Literal & \cs{frac}    \\
      inter-unit-product           & Literal & \cs{,}       \\
      parse-units                  & Switch  & true         \\
      per-mode                     & Choice  & power        \\
      per-symbol-script-correction & Literal & \cs{!}       \\
      per-symbol                   & Literal & /            \\
      power-half-as-sqrt           & Switch  & false        \\
      qualifier-mode               & Choice  & subscript    \\
      qualifier-phrase             & Literal & \meta{empty} \\
      sticky-per                   & Switch  & false        \\
      unit-font-command            & Literal & \cs{mathrm}  \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{inter-unit-product}
The separator between each unit is stored using the \opt{inter-unit-product}
option. The standard setting is a thin space: another common choice is a
centred dot. To get the correct spacing it is necessary to use
|\ensuremath{{}\cdot{}}| in the latter case.
\begin{LaTeXdemo}
  \unit{\farad\squared\lumen\candela} \\
  \unit[inter-unit-product = \ensuremath{{}\cdot{}}]
    {\farad\squared\lumen\candela}
\end{LaTeXdemo}

\DescribeOption{per-mode}
\DescribeOption{display-per-mode}
\DescribeOption{inline-per-mode}
\DescribeOption{per-symbol}
\DescribeOption{fraction-command}
\DescribeOption{bracket-unit-denominator}
The handling of \cs{per} is altered using the \opt{per-mode} choice option. The
standard setting is \opt{power}, meaning that \cs{per} generates reciprocal
powers for units. Setting the option to \opt{fraction} uses the |\frac|
function to typeset the positive and negative powers of a unit separately.
The exact function can be adjusted using the \opt{fraction-command} option.
\begin{LaTeXdemo}
  \unit{\joule\per\mole\per\kelvin} \\
  \unit{\metre\per\second\squared} \\
  \unit[per-mode = fraction]{\joule\per\mole\per\kelvin} \\
  \unit[per-mode = fraction]{\metre\per\second\squared}
\end{LaTeXdemo}
The closely-related \opt{power-positive-first} setting acts in the same way but
places all of the positive powers before any negative ones.
\begin{LaTeXdemo}
  \unit{\ampere\per\mole\second} \\
  \unit[per-mode = power-positive-first]
    {\ampere\per\mole\second}
\end{LaTeXdemo}
It is possible to use a symbol (usually |/|) to separate the two parts of a
unit by setting \opt{per-mode} to \opt{symbol}; the symbol used is stored using
the setting \opt{per-symbol}. This method for displaying units can be
ambiguous, and so brackets are added unless \opt{bracket-unit-denominator} is
set to \opt{false}. Notice that \opt{bracket-unit-denominator} only applies
when \opt{per-mode} is set to \opt{symbol}.
\begin{LaTeXdemo}
  \sisetup{per-mode = symbol}%
  \unit{\joule\per\mole\per\kelvin} \\
  \unit{\metre\per\second\squared} \\
  \unit[per-symbol = \ \text{div}\ ]{\joule\per\mole\per\kelvin} \\
  \unit[bracket-unit-denominator = false]{\joule\per\mole\per\kelvin}
\end{LaTeXdemo}
The often-requested (but mathematically invalid) \opt{repeated-symbol} option
is also available to repeat the symbol for each \cs{per}.
\begin{LaTeXdemo}
  \unit[per-mode = repeated-symbol]{\joule\per\mole\per\kelvin}
\end{LaTeXdemo}
The use of a symbol can be restricted to the case where exactly one is
required: the setting \opt{single-symbol}  will use a symbol if and only
if there are one or more positive powers and exactly one negative power.
In other cases, powers are used.
\begin{LaTeXdemo}
  \sisetup{per-mode = single-symbol}
  \qty{10}{\per\metre} \\
  \qty{20}{\metre\per\second} \\
  \qty{30}{\joule\per\mole\per\kelvin}
\end{LaTeXdemo}
It is possible for the behaviour of the \cs{per} function to depend on the
prevailing math style. Setting either \opt{display-per-mode} or
\opt{inline-per-mode} independently can be used to achieve this. For example,
the following example will will use the \opt{symbol} setting for in line math,
and the \opt{fraction} setting when used in display math.
\begin{LaTeXdemo}
  \sisetup{
    display-per-mode = fraction ,
    inline-per-mode = symbol
  }%
  $ \unit{\joule\per\mole\per\kelvin} $
  \[ \unit{\joule\per\mole\per\kelvin} \]
     \unit{\joule\per\mole\per\kelvin} \\
  $
    \displaystyle
    \unit{\joule\per\mole\per\kelvin}
  $
  \[
    \textstyle
    \unit{\joule\per\mole\per\kelvin}
  \]
\end{LaTeXdemo}

\DescribeOption{per-symbol-script-correction}
When using the \opt{symbol} setting for \opt{per-mode}, there may the need to
adjust spacing between a superscript power and the symbol. This is provided as
a command to be inserted between the two items: the standard value is a thin
negative space, \cs{!}.
\begin{LaTeXdemo}
  \sisetup{per-mode = symbol}%
  \unit{\cm\cubed\per\gram} \\
  \unit[per-symbol-script-correction = ]{\cm\cubed\per\gram}
\end{LaTeXdemo}

\DescribeOption{sticky-per}
By default, \cs{per} applies only to the next unit given.\footnote{This is the
standard method of reading units in English: for example,
\unit{\joule\per\mole\per\kelvin} is pronounced \enquote{joules per mole per
kelvin}.} By setting the \opt{sticky-per} flag, this behaviour is changed so
that \cs{per} applies to all subsequent units.
\begin{LaTeXdemo}
  \unit{\pascal\per\gray\henry} \\
  \unit[sticky-per]{\pascal\per\gray\henry}
\end{LaTeXdemo}

\DescribeOption{qualifier-mode}
\DescribeOption{qualifier-phrase}
Unit qualifiers can be printed in three different formats, set by the
\opt{qualifier-mode} option. The standard setting is \opt{subscript}, while the
options \opt{bracket}, \opt{combine} and \opt{phrase} are also possible. With
the last settings, powers can lead to ambiguity and are automatically detected
and brackets added as appropriate.
\begin{LaTeXdemo}
  \unit{\kilogram\of{pol}\squared\per\mole\of{cat}\per\hour} \\
  \unit[qualifier-mode = bracket]
    {\kilogram\of{pol}\squared\per\mole\of{cat}\per\hour} \\
  \unit[qualifier-mode = combine]
    {\deci\bel\of{i}}
\end{LaTeXdemo}
The \opt{phrase} option is used with \opt{qualifier-phrase}, which allows for
example a space or other linking text to be inserted.
\begin{LaTeXdemo}
  \sisetup{qualifier-mode = phrase, qualifier-phrase = \ }%
  \unit{\kilogram\of{pol}\squared\per\mole\of{cat}\per\hour} \\
  \sisetup{qualifier-phrase = \ \mbox{of}\ }%
  \unit{\kilogram\of{pol}\squared\per\mole\of{cat}\per\hour}
\end{LaTeXdemo}

\DescribeOption{power-half-as-sqrt}
In some cases, the power of $0.5$ is shown by giving the unit symbol as a
square root. This can be enabled by setting \opt{power-half-as-sqrt} to
\opt{true}
\begin{LaTeXdemo}
  \unit{\Hz\tothe{0.5}} \\
  \unit[power-half-as-sqrt]{\Hz\tothe{0.5}}
\end{LaTeXdemo}

\DescribeOption{parse-units}
Normally, \pkg{siunitx} is used with the unit parse enabled, and only prints
units directly if there is literal input. However, if the input is known to be
essentially consistent and high performance is desired, then the parser can be
turned off using the \opt{parse-units} switch.
\begin{LaTeXdemo}
  \qty{300}{\MHz} \\
  \qty[parse-units = false]{300}{\MHz}
\end{LaTeXdemo}

\DescribeOption{forbid-literal-units}
Some users may prefer to completely disable the use of literal input in units,
for example to enforce consistency. This can be accomplished by setting the
\opt{forbid-literal-units} switch. With this option enabled, only macro-based
units can be used in a document. This only applies when \opt{parse-units} is
\opt{true}.

\DescribeOption{unit-font-command}
The command used to set unit themselves may be adjusted using the
\opt{unit-font-command} option. This is typically set to |\mathrm|.
\begin{LaTeXdemo}
  \unit{\lumen} \\
  \unit[unit-font-command = \mathit]{\lumen}
\end{LaTeXdemo}

\subsection{Quantities}

Some options apply to quantities (the combination of a number and a unit),
rather than to the numbers or units alone (Table~\ref{tab:opt:quantities}).
\begin{table}
  \centering
  \caption{Options for quantities.%
    \label{tab:opt:quantities}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{@{}l}{Default} \\
    \midrule
      allow-quantity-breaks      & Switch  & false   \\
      extract-mass-in-kilograms  & Switch  & true    \\
      prefix-mode                & Choice  & input   \\
      quantity-product           & Literal & \cs{,}  \\
      separate-uncertainty-units & Choice  & bracket \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{allow-quantity-breaks}
Usually, the combination of a number and unit is regarded as a single
mathematical entity which should not be split across lines. However, there are
cases (very long units, narrow columns, \foreign{etc}.) where breaks may be
needed. This can be turned on using the \opt{allow-quantity-breaks} option.
\begin{LaTeXdemo}
  \begin{minipage}{2.55cm}
    % Gives an underfull hbox
    Some filler text \qty{10}{\metre} \\
    \sisetup{allow-quantity-breaks}
    Some filler text \qty{10}{\metre}
  \end{minipage}
\end{LaTeXdemo}

\DescribeOption{quantity-product}
The product symbol between the number and unit is set using the
\opt{quantity-product} option.
\begin{LaTeXdemo}
  \qty{2.67}{\farad} \\
  \qty[quantity-product = \ ]{2.67}{\farad} \\
  \qty[quantity-product = ]{2.67}{\farad}
\end{LaTeXdemo}

\DescribeOption{prefix-mode}
\DescribeOption{extract-mass-in-kilograms}
The unit prefixes (\cs{kilo}, \foreign{etc}.) are normally given as letters.
However, the package can convert these into numerical powers. This is
controlled by the \opt{prefix-mode} option, which takes the values \opt{input},
\opt{combine-exponent} and \opt{extract-exponent}. The \opt{input} setting
leaves the prefixes unchanged. Using \opt{combine-exponent} will add any
exponent amount from the number to the first unit: this will modify any
existing prefix. Finally, using \opt{extract-exponent} will remove all prefixes
and express them as an exponent. The treatment of kilograms in this case can be
set using \opt{extract-mass-in-kilograms}: when \opt{true}, the \emph{kilo}
prefix is retained as part of the unit. This will mean that all grams are
converted to kilograms.
\begin{LaTeXdemo}
  \qty{1e3}{\metre\second} \\
  \qty[prefix-mode = combine-exponent]{1e3}{\metre\second} \\
  \qty{10}{\kilo\gram\squared\deci\second} \\
  \qty[prefix-mode = extract-exponent]{10}{\kilo\gram\squared\deci\second}\\
  \qty[prefix-mode = extract-exponent]{7.5}{\gram} \\
  \sisetup{extract-mass-in-kilograms = false}
  \qty{10}{\kilo\gram\squared\deci\second} \\
  \qty[prefix-mode = extract-exponent]{10}{\kilo\gram\squared\deci\second} \\
  \qty[prefix-mode = extract-exponent]{7.5}{\gram} \\
\end{LaTeXdemo}

\DescribeOption{separate-uncertainty-units}
When a number has multiple parts (such as a separate uncertainty) then the unit
must apply to all parts of the number. How this is shown is controlled using
the \opt{separate-uncertainty-units} options. The standard setting is
\opt{brackets}, which will place the entire numerical part in brackets and use
a single unit symbol. Alternative options are \opt{repeat} (print the unit for
each part of the number) and \opt{single} (print only one unit symbol:
mathematically incorrect).
\begin{LaTeXdemo}
  \sisetup{uncertainty-mode = separate}
  \qty{12.3(4)}{\kilo\gram} \\
  \qty[separate-uncertainty-units = bracket]{12.3(4)}{\kilo\gram} \\
  \qty[separate-uncertainty-units = repeat]{12.3(4)}{\kilo\gram}  \\
  \qty[separate-uncertainty-units = single]{12.3(4)}{\kilo\gram}
\end{LaTeXdemo}

\subsection{Tabular material}

Processing of material in tables obeys the same settings as described for the
functions already described. However, there are some settings which apply only
to the layout of tabular material (Table~\ref{tab:opt:tables}).
\begin{table}
  \caption{Options for tabular material.%
    \label{tab:opt:tables}}
  \begin{tabular}{@{}>{\ttfamily}ll>{\ttfamily}l@{}}
    \toprule
      \multicolumn{1}{@{}l}{Option name} &
      Type &
      \multicolumn{1}{l@{}}{Default} \\
    \midrule
      table-align-comparator    & Switch  & true         \\
      table-align-exponent      & Switch  & true         \\
      table-align-text-after    & Switch  & true         \\
      table-align-text-before   & Switch  & true         \\
      table-align-uncertainty   & Switch  & true         \\
      table-alignment           & Meta    & center       \\
      table-alignment-mode      & Choice  & marker       \\
      table-auto-round          & Switch  & false        \\
      table-column-width        & Length  & 0pt          \\
      table-fixed-width         & Switch  & false        \\
      table-format              & Special & 2.2          \\
      table-model-setup         & Literal & \meta{empty} \\
      table-number-alignment    & Choice  & center       \\
      table-text-alignment      & Choice  & center       \\
    \bottomrule
  \end{tabular}
\end{table}

\DescribeOption{table-alignment-mode}
The method used by \pkg{siunitx} to align numbers is selected using the
\opt{table-alignment-mode} option, which may be one of \opt{marker},
\opt{format} or \meta{none}. With the standard setting, \opt{marker}, the
package centers the decimal marker in a tabular column, potentially leaving
white space at the shorter end of a number. The \opt{format} mode uses
information from the \opt{table-format} key to construct a model: this is then
used to define the space available to a number. For asymmetrical numbers, this
method is strongly preferable. Finally, \opt{none} disables alignment entirely:
numbers are simply parsed.

\DescribeOption{table-number-alignment}
When \opt{table-alignment-mode} is set to \opt{format} or \opt{none}, the
placement of the number \enquote{block} within the cell as a whole is set by
the \opt{table-number-alignment} option, which may be one of \opt{left},
\opt{center} or \opt{right}. (When \opt{table-alignment-mode} is set to
\opt{marker}, the decimal marker is always centered in the cell.) The
different alignment choices are illustrated in Table~\ref{tab:S:align}, which
uses somewhat exaggerated column headings to show the relative position of the
cell contents.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Aligning the \texttt{S} column.%
      \label{tab:S:align}}
    \centering
    \sisetup{table-format = 2.4, table-alignment-mode = format}
    \begin{tabular}{@{}
      S[table-alignment-mode = marker]
      S[table-number-alignment = center]
      S[table-number-alignment = left]
      S[table-number-alignment = right]
    @{}}
    \toprule
    {Some Values} & {Some Values} & {Some Values} & {Some Values} \\
    \midrule
       2.3456  &  2.3456  &  2.3456 &  2.3456 \\
      34.2345  & 34.2345  & 34.2345 & 34.2345 \\
      56.7835  & 56.7835  & 56.7835 & 56.7835 \\
      90.473   & 90.473   & 90.473  & 90.473  \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}
When the alignment mode is set to \opt{none}, number are simply collected and
parsed without any further processing, as illustrated in
Table~\ref{tab:S:parse}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Parsing without aligning in an \texttt{S} column.%
      \label{tab:S:parse}}
    \begin{tabular}
      {@{}
        S
        S[table-alignment-mode = none]
      @{}}
      \toprule
        {Decimal-centred} &
        {Simple centring} \\
      \midrule
        12.345   & 12.345   \\
        6,78     & 6,78     \\
        -88.8(9) & -88.8(9) \\
        4.5e3    & 4.5e3    \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-format}
When the \opt{table-alignment-mode} is set to \opt{format}, \pkg{siunitx} uses
the information set in \opt{table-format} to construct a \enquote{model} which
defines the space to reserve for a number. The \opt{table-format} key is
interpreted in much the same way as a table cell. The numerical part should
consist of a number showing how many figures to reserve in each part of the
input, plus any comparators, signs, \foreign{etc.} A variety of examples are
given in Table~\ref{tab:S:format}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Reserving space in \texttt{S} columns.%
      \label{tab:S:format}}
    \sisetup{
      table-alignment-mode   = format,
      table-number-alignment = center,
    }
    \begin{tabular}{@{}
      S[table-format = 2.2]
      S[table-format = 2.2, table-number-alignment = right]
      S[table-format = 2.2(1)]
      S[table-format = 2.2(1), uncertainty-mode = separate]
      S[table-format = +2.2]
      S[table-format = 2.2e1]
     @{}}
    \toprule
        {Values}
      & {Values}
      & {Values}
      & {Values}
      & {Values}
      & {Values} \\
    \midrule
       2.3  &  2.3  &  2.3(5)  &  2.3(5)  &   2.3  &  2.3e8  \\
      34.23 & 34.23 & 34.23(4) & 34.23(4) &  34.23 & 34.23   \\
      56.78 & 56.78 & 56.78(3) & 56.78(3) & -56.78 & 56.78e3 \\
       3,76 &  3,76 &  3,76(2) &  3.76(2) & +-3.76 &      e6 \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}
It is important to note that any parts of a number \emph{not} specified in the
table format argument are set to be absent (the number of figures is set to
zero). Setting the \opt{table-format} option also resets
\opt{table-alignment-mode} to \opt{format}.

\DescribeOption{table-model-setup}
In some tables (particularly those using bold extended fonts), it may be
necessary to insert additional information when creating the \enquote{model}.
This is handled using the \opt{table-model-setup} key. An example of the use of
this key is given in Section~\ref{sec:hint:heading}.

Space for material before and after the \texttt{S} column can be reserved by
giving model text as part of the \opt{table-format} key. This is then used to
provide the necessary gap while maintaining alignment (Table~\ref{tab:S:ends}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Text before and after numbers.%
      \label{tab:S:ends}}
    \sisetup{table-format = {now }2.4{\textsuperscript{\emph{a}}}}
    \begin{tabular}{@{}S@{}}
    \toprule
    {Values} \\
    \midrule
            2.3456  \\
           34.2345  \textsuperscript{\emph{a}}\\
           56.7835  \\
      now~ 90.473   \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-align-exponent}
\DescribeOption{table-align-uncertainty}
\DescribeOption{table-align-comparator}
When printing exponents in tables, there is a choice of aligning the exponent
parts or having these close up to the mantissa. This is controlled by the
\opt{table-align-exponent} option (Table~\ref{tab:align:exp}). Similarly,
uncertainty parts which are printed separately from the mantissa can be aligned
or closed up. This is set by the \opt{table-align-uncertainty} option
(Table~\ref{tab:align:uncert}). Finally, the same approach is available for the
comparator with the \opt{table-align-comparator} option
(Table~\ref{tab:align:comp}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{The \opt{table-align-exponent} option.%
      \label{tab:align:exp}}
    \sisetup{table-format = 1.3e2}
    \begin{tabular}{@{}SS[table-align-exponent = false]@{}}
      \toprule
        {Header} & {Header} \\
      \midrule
        1.2e3    & 1.2e3    \\
        1.234e56 & 1.234e56 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{The \opt{table-align-uncertainty} option.%
      \label{tab:align:uncert}}
    \sisetup{
      uncertainty-mode = separate,
      table-format = 1.3(1),
    }
    \begin{tabular}{@{}SS[table-align-uncertainty = false]@{}}
      \toprule
        {Header} & {Header} \\
      \midrule
        1.2(1)   & 1.2(3)    \\
        1.234(5) & 1.234(5) \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{The \opt{table-align-comparator} option.%
      \label{tab:align:comp}}
    \sisetup{table-format = >2.2}
    \begin{tabular}{@{}SS[table-align-comparator = false]@{}}
      \toprule
        {Header} & {Header} \\
      \midrule
        >  1.2  & >  1.2  \\
        < 12.34 & < 12.34 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-align-text-before}
\DescribeOption{table-align-text-after}
Note markers are often given in tables after the numerical content. It may be
desirable for these to close up to the numbers. Whether this takes place is
controlled by the \opt{table-align-text-before} and \opt{\ldots-after}
 option (Table~\ref{tab:S:notes}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Closing notes up to text.%
      \label{tab:S:notes}}
    \newrobustcmd\NoteMark[1]{%
      \textsuperscript{\emph{#1}}%
    }
    \sisetup{table-format = {\NoteMark{a}}2.4}
    \begin{tabular}{@{}
      S
      S[table-align-text-before = false]
    @{}}
    \toprule
      {Values}             & {Values} \\
    \midrule
                    2.3456 &               2.3456 \\
      \NoteMark{a}  4.234  & \NoteMark{a}  4.234 \\
      \NoteMark{b}   .78   & \NoteMark{b}   .78  \\
      \NoteMark{d} 88      & \NoteMark{d} 88     \\
    \bottomrule
    \end{tabular}
    \hspace{\fill}%
    \sisetup{table-format = 2.4\NoteMark{a}}
    \begin{tabular}{@{}
      S
      S[table-align-text-after = false]
    @{}}
    \toprule
      {Values}             & {Values} \\
    \midrule
       2.3456              & 2.3456 \\
      34.234  \NoteMark{a} & 34.234 \NoteMark{a} \\
      56.78   \NoteMark{b} & 56.78  \NoteMark{b} \\
      90.4    \NoteMark{c} & 90.4   \NoteMark{c} \\
      88      \NoteMark{d} & 88     \NoteMark{d} \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-auto-round}
The contents of table cells can automatically be rounded or zero-filled to the
number of decimal digits given for the decimal part of the \opt{table-format}
option. This mode is activated using the \opt{table-auto-round} switch, as
illustrated in Table~\ref{tab:S:auto}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \centering
    \caption{The \opt{table-auto-round} option.%
      \label{tab:S:auto}}
    \sisetup{table-format = 1.3}
    \begin{tabular}{@{}SS[table-auto-round]@{}}
      \toprule
        {Header} & {Header} \\
      \midrule
        1.2    & 1.2    \\
        1.2345 & 1.2345 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{parse-numbers}
When the \opt{parse-numbers} option is set to \opt{false}, then the alignment
code for tables takes a different approach. The output is always set in math
mode, and alignment takes place at the first decimal marker. This is achieved
by making it active in math mode. When reserving space for content only the
integer and decimal values for the mantissa are considered
(Table~\ref{tab:S:nonparsed}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Aligning without parsing.%
      \label{tab:S:nonparsed}}
    \sisetup{
      parse-numbers = false,
      table-format  = 3.3
    }
    \centering
    \begin{tabular}{@{}
      S
      S[table-number-alignment = center]
      S[table-number-alignment = right]
      S[table-number-alignment = left]
      @{}}
    \toprule
        {Some values}
      & {Some values}
      & {Some values}
      & {Some values} \\
    \midrule
       2.35  &    2.35  &    2.35  &   2.35   \\
      34.234 &   34.234 &   34.234 &  34.234  \\
      56.783 &   56.783 &   56.783 &  56.783  \\
       3,762 &    3,762 &    3,762 &   3.762  \\
    \sqrt{2} & \sqrt{2} & \sqrt{2} & \sqrt{2} \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{drop-exponent}
In cases where data cover a range of values, printing using a fixed exponent in
a table may make presentation clearer. In these cases, dropping the exponent
value from the table is useful. The general numerical options \opt{drop-exponent}
combined with \opt{exponent-mode = fixed} can be used to achieve this
(Table~\ref{tab:exp:omit}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{The \opt{drop-exponent} option.%
      \label{tab:exp:omit}}
    \begin{tabular}{@{}
      S[table-format = 1.1e1]
      S[
        drop-exponent  = true,
        exponent-mode  = fixed,
        fixed-exponent = 3,
        table-format   = 2.1,
      ]
    @{}}
      \toprule
        {Header} &
          \multicolumn{1}{c@{}}{Header / \num[print-unity-mantissa = false]{e3}} \\
      \midrule
        1.2e3 & 1.2e3 \\
         3e2  &   3e2 \\
       1.0e4  & 1.0e4 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-column-width}
\DescribeOption{table-fixed-width}
Usually, the width of the numerical column is allowed to vary depending on the
content. However, there are cases where a strictly fixed width is desirable.
For these cases, the \opt{table-fixed-width} and \opt{table-column-width}
options are available. The \opt{table-fixed-width} option activates fixed-width
columns, whilst \opt{table-column-width} defines the target width
(Table~\ref{tab:width:fixed}). Setting \opt{table-column-width} to a positive
value automatically enables \opt{table-fixed-width}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Fixed-width columns.%
      \label{tab:width:fixed}}
    \begin{tabular}
      {@{}
        S
        S[table-column-width = 2cm]
      @{}}
      \toprule
        {Flexible} &
        {Fixed}    \\
      \midrule
        1.23 & 1.23 \\
        45.6 & 45.6 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

The \opt{table-column-width} option can also be used to achieve special
effects. One example is centring a column of numbers under a wide heading, with
the numbers themselves right-aligned (Table~\ref{tab:width:special}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \centering
    \caption{Right-aligning under a heading.%
      \label{tab:width:special}}
    \settowidth{\mylength}{Long header}
    \sisetup{
      table-alignment-mode   = none      ,
      table-column-width     = \mylength ,
      table-number-alignment = right
    }
    \begin{tabular}{@{}S@{}}
    \toprule
    {Long header} \\
    \midrule
        12.33 \\
         2    \\
      1234    \\
     \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-text-alignment}
\DescribeOption{table-alignment}
Cell contents which are not part of a number can be protected using braces, as
illustrated. Cells which contain no numerical data at all are aligned using the
setting specified by the \opt{table-text-alignment} option, which recognises
the values \opt{center}, \opt{left}, \opt{none} and \opt{right}
(Table~\ref{tab:S:text}). The setting \opt{none} is intended for use with
the package \pkg{tabularray}, which carries out its own alignment of textual
values.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Aligning text in \texttt{S} columns.%
      \label{tab:S:text}}
    \sisetup{table-format = 4.4}
    \centering
    \begin{tabular}{@{}
      S
      S[table-text-alignment = left]
      S[table-text-alignment = right]
    @{}}
    \toprule
        {Values}
      & {Values}
      & {Values} \\
    \midrule
      992.435  &  992.435  &  992.435  \\
     7734.2344 & 7734.2344 & 7734.2344 \\
       56.7834 &   56.7834 &   56.7834 \\
        3,7462 &    3,7462 &    3,7462 \\
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\DescribeOption{table-alignment}
The table alignment options \opt{table-number-alignment} and
\opt{table-text-alignment} can be set to the same value using the
\opt{table-alignment} option. This will set all three alignment options to the
same value (one of \opt{center}, \opt{right} or \opt{left}).

\subsection{Locale options}

\DescribeOption{locale}
\pkg{siunitx} allows the user to switch between the typographic conventions of
different (geographical) areas by using locales. Currently, the package is
supplied with configurations for locales \opt{UK}, \opt{US}, \opt{DE}
(Germany), \opt{IT} (Italian) \opt{PL} (Poland), \opt{FR} (French), \opt{SI}
(Slovene) and \opt{ZA} (South Africa). The \opt{locale} option is used to
switch to a particular locale.
\begin{LaTeXdemo}
  \qty{1.234}{\metre}\\
  \qty[locale = DE]{6.789}{\metre}
\end{LaTeXdemo}

\subsection{Preamble-only options}

\DescribeOption{list-input-separator}
\DescribeOption{product-input-separator}
The separator between \emph{input} elements of lists (\cs{numlist},
\cs{qtylist}) and products (\cs{numproduct}, \cs{qtyproduct}) can be adjusted
using the options \opt{list-input-separator} and \opt{product-input-separator},
respectively. These are load-time only options as the separator here is part of
the \enquote{document interface} rather than part of the individual number(s).
The supplied value should be a \emph{single} marker.

\DescribeOption{table-column-type}
The letter(s) used to create table columns can be adjusted using the
\opt{table-column-type} option. The standard setting is |S|, but one or more
letters may be used: these must be single tokens. For example, provided the
\pkg{numprint} package has not been loaded, the letter |n| could be used as
this would suggest a numerical column.

\section{Significant new features updates}

Here, a short reminder for users of new features added since the release of
version~\version{3} of the package is given.

\subsection{Version~\version{3.1}}

\begin{itemize}
  \item Extend support for complex numbers to cover polar form including
    new options to control output
  \item Parse and format multiple uncertainties; new options added
    to allow descriptions of each uncertainty
  \item Customisation of text sub/superscript output
  \item Selectable group size when dividing digits
  \item Finer controls for \enquote{per mode} output
  \item Support for negative zero values and controlled output for zero
    values
  \item Allow negative values to round to zero
  \item Addition unit abbreviations for capacitance, power and magnetic
    flux
  \item Portuguese locale
\end{itemize}

\subsection{Version~\version{3.2}}

\begin{itemize}
  \item \enquote{Threshold} approach to producing exponent notation:
    similar to a calculator
  \item Finer control of printing plus signs using new options to split
    mantissa and exponent
  \item Support for updated prefixes introduced by \acro{bipm}
  \item Revised alignment of uncertainties in tabulars
\end{itemize}

\subsection{Version~\version{3.3}}

\begin{itemize}
  \item Support for languages which need a begin as well as end phrase in
    ranges, \foreign{e.g.}~Italian, was added; this includes appropriate
    supporting options
  \item Support for asymmetric uncertainties (tolerances)
  \item Rounding extended to allow truncation (rounding down) for both
    main values and when using the uncertainty part
  \item Finer control of the appearance of brackets in ambiguous
     output introduced by adding options for different contexts
  \item Control of list and product input and output extended
  \item Finer control of table alignment setup for bold values
  \item New unit abbreviations for magnetic fields
\end{itemize}

\subsection{Version~\version{3.4}}

\begin{itemize}
  \item Support uncertainties in seconds part of arc angles
  \item Control creation of minute and second components on conversion of
    integer angles to arc format
  \item Ability to simplify \enquote{asymmetric} uncertainties with
    equal components in both directions
\end{itemize}

\section{Upgrading from version~\version{2}%
  \label{sec:upgrading}}

The package has been largely re-written internally between versions \version{2}
and \version{3}. A significant number of key--value settings have new, more
descriptive, names. Where possible, older names are mapped to newer ones
internally: you will be warned in the log if this is the case.

It is possible to use the \LaTeXe{} kernel mechanism to load the last
version~\version{2} release for documents that cannot be successfully processed
using version~\version{3}. This can be achieved using
\begin{LaTeXdemo}[code only]
  \usepackage{siunitx}[=2021-04-09]
\end{LaTeXdemo}
or
\begin{LaTeXdemo}[code only]
  \usepackage{siunitx}[=v2]
\end{LaTeXdemo}
This approach will work with older systems which still have version~\version{2}
installed, meaning that you can reliably use it to work between systems with
different versions of \pkg{siunitx}.

\begin{function}{\SI, \SIlist, \SIrange, \si}
  In version~\version{3}, the document commands have been revised to be more
  descriptive. As such, the commands \cs{SI}, \cs{SIlist}, \cs{SIrange} and
  \cs{si} remain available but are not recommended for use in new documents.
  Use the new \cs{qty\ldots} commands instead: they are clearer and in some
  cases very slightly faster.
\end{function}

Some changes have been made to the semantics of commands or options. Most
notably
\begin{itemize}
  \item prefixes cannot now be given without units;
  \item prefixes can only be interconverted with numbers as part
    of a quantity, not as stand-alone units.
\end{itemize}
See Section~\ref{sec:hint:prefixes} for how to work with the new approach
if you want to print prefix information.

The font control system has been completely re-written for version~\version{3}.
The method used is entirely different from version~\version{2}. Emulation is
therefore not provided for all outcomes: if you need non-standard font
settings, you will need to adjust your source. See Section~\ref{sec:print} for
more details on the options available in this area.

The input approach for version~\version{3} is slightly more structured and
restricted than for version~\version{2}. As well as the updated names for
document commands, this means that
\begin{itemize}
  \item Products of numbers must now be given using the dedicated
    \cs{numproduct} and \cs{qtyproduct} commands;
  \item Quotients of numbers are only supported as literals;
  \item Complex values need to be given using the dedicated command
    \cs{complexnum}.
\end{itemize}
The option \opt{round-integer-to-decimal = false} has been removed, and whilst
there is not a direct replacement, users are likely to find that \opt{round-pad
= false} offers the desired outcome.

A new approach has been taken to providing non-Latin symbols for use in
units: these are now handled directly where needed, for example in the
definition of the \cs{micro} prefix.

Translation of fixed strings is now carried out using the \pkg{translations}
package. If you have manually set up translations in version~\version{2} using
\pkg{translator}, you will need to load it manually.

The letter used for a numerical tabular column can now be selected by the
user: the letter |S| has been retained as the standard interface. The unit
column (|s|) has been removed from this release. It can be emulated using
the \pkg{collcell} package, for example
\begin{LaTeXdemo}[code only]
  \usepackage{collcell}
  \newcolumntype{s}{>{\collectcell\unit}c<{\endcollectcell}}
\end{LaTeXdemo}
or
\begin{LaTeXdemo}[code only]
  \usepackage{collcell}
  \newcolumntype{s}{>{\collectcell\si}c<{\endcollectcell}}
\end{LaTeXdemo}

If you are using \opt{table-column-width} to have fixed-width columns, you
also now need \opt{table-fixed-width} to set this option active.

Direct support for loading a local configuration file, \texttt{siunitx.cfg},
has been removed. However, the approach described in
Section~\ref{sec:hint:config} may be used to achieve the same effect with the
additions more clearly shown in document sources.

The command \cs{SendSettingsToPgf} is deprecated, and should be replaced
simply by setting the appropriate \cs{pgfkeys} in parallel to \cs{sisetup}.

\section{Unit changes made by \acro{BIPM}%
  \label{sec:BIPM8}}

In addition to the changes in the \pkg{siunitx} package described in
Section~\ref{sec:upgrading}, there are changes in the units defined by the
\acro{BIPM} in the 9th Edition of the \acro{SI} Brochure which are reflected
here. There are two major areas of change.

The first is in respect of units accepted for use with \acro{SI} units. In the
8th Edition of the \acro{SI} Brochure, the following units were listed as
accepted for use in specialist fields
\begin{itemize}
  \item ångström (\cs{angstrom})
  \item bar (\cs{bar})
  \item barn (\cs{barn})
  \item knot (\cs{knot})
  \item millimetre of mercury (\cs{mmHg})
  \item nautical mile (\cs{nauticalmile})
\end{itemize}
These are no longer listed in the 9th Edition, and so are deprecated as
pre-defined units by \pkg{siunitx}. These units will issue a warning on first
use, and users should add their own definitions to the start of their sources
to avoid this.

Secondly, the move to a new definition of base units means that the table of
units determined experimentally has been removed from the \acro{SI} Brochure.
This covers the following units defined by \pkg{siunitx} in previous releases
\begin{itemize}
  \item \cs{bohr}
  \item \cs{clight}
  \item \cs{electronmass}
  \item \cs{elementarycharge}
  \item 	\cs{hartree}
  \item \cs{planckbar}
\end{itemize}
These are also deprecated in \pkg{siunitx} and users should provide their own
definitions.

In addition to these two major blocks, the unit \cs{atomicmassunit} has similar
deprecated status: this was listed as with experimentally-determined units in
the 8th Edition of the \acro{SI} Brochure but is equivalent to the dalton, a
unit which remains accepted.

\section{Localisation}

The \pkg{translations} package provides a structured framework for
localisation of words and phrases. In particular, it offers the
\cs{GetTranslation} macro, which will provide appropriate translations based
on the current \pkg{babel} or \pkg{polyglossia} language setting.

If \pkg{translations} is available, \pkg{siunitx} will load it and alter the
standard settings for the \opt{list-final-separator} and \opt{range-phrase}
options to read:
\begin{LaTeXdemo}[code only]
  \sisetup{
    list-final-separator = { \GetTranslation{and} },
    list-pair-separator  = { \GetTranslation{and} },
    range-phrase         = { \GetTranslation{to (numerical range)} },
  }
\end{LaTeXdemo}
The setting for \opt{range-open-phrase} is also adjusted to support a setting
for \opt{range-open-phrase}: this requires some conditionals so is more
complex.

If the current language is known to the \pkg{translations} package then the
result will be localised text. The preamble for this manual loads English,
French, German, Italian, Polish, Spanish, Catalan, Portuguese and Brazilian as
options, and also loads the \pkg{babel} package:
\begin{LaTeXdemo}
  % In English by default
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{french}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{german}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{italian}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{polish}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{spanish}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{catalan}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{portuguese}%
  \numlist{1;2;3} \\
  \numrange{1}{10} \\
  \selectlanguage{brazilian}%
  \numlist{1;2;3} \\
  \numrange{1}{10}
\end{LaTeXdemo}

\section{Compatibility with other packages}

In general, \pkg{siunitx} should be usable with other packages without
interference.

When the \pkg{physics} package is loaded before \pkg{siunitx}, the command
\cs{qty} is not defined. Users may use the version~\version{2} command \cs{SI},
which can be used as a drop-in replacement for \cs{qty}. Alternatively, if the
\pkg{siunitx} definition is preferred, you may use
\begin{LaTeXdemo}[code only]
  \AtBeginDocument{\RenewCommandCopy\qty\SI}
\end{LaTeXdemo}
and use the longer name \cs{quantity} to access the functionality of the
\pkg{physics} package.

\section{Hints for using \pkg{siunitx}}

\subsection{Problematic font encodings%
  \label{sec:hint:encoding}}

The standard settings in \pkg{siunitx} assume that \enquote{sensible} input and
font encoding values prevail. The input encoding is assumed to be \acro{UTF-8}
in all cases. With pdf\LaTeX{}, the font encoding should be \texttt{T1},
whereas for Xe\LaTeX{} and Lua\LaTeX{}, \texttt{TU} (Unicode font encoding) is
expected.

Some packages, for example \pkg{newtxtext} or \pkg{stix2}, either force
\texttt{T1} or do not anticipate \texttt{TU} correctly with Xe\LaTeX{} and
Lua\LaTeX{}. In these cases, the symbols used by \pkg{siunitx} may be
incorrect. If correcting the font encoding is not possible, you will need to
re-declare the relevant units using symbol definitions which account for this
non-standard setup.

\subsection{Adjusting \cs{litre} and \cs{liter}%
  \label{sec:hint:litre}}

As detailed earlier, the unit macros \cs{litre} and \cs{liter} are both
available for litres. With the standard settings, \cs{liter} is defined as
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\liter{\litre}
\end{LaTeXdemo}
meaning that \cs{litre} is the \enquote{canonical} unit. This follows the same
relationship as exists between \cs{metre} and \cs{meter}.

In contrast to metres, however, there is more likelihood of users wishing to
adjust the appearance of litres: both \enquote{\unit{l}} and \enquote{\unit{L}}
are commonly used. The recommended approach to adjustment is to re-declare
the \cs{litre} macro, as \cs{liter} will follow automatically.
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\litre{l}
\end{LaTeXdemo}

\subsection{Ensuring text or math output%
  \label{sec:hint:text-math}}

The macros \cs{ensuremath} and \cs{text} should be used to ensure that a
particular item is always printed in the desired mode. Some mathematical output
does not work well in \cs{mathrm} (the font setting used by \pkg{siunitx} for
printing units). The easiest way to solve this is to use the construction
"\text{\ensuremath{...}}", which will print the material in the standard
mathematics font without affecting the rest of the output. In some cases,
simply forcing \cs{mathnormal} will suffice, but this is less reliable with
non-Latin characters.

\subsection{Including a literal hyphen inside \cs{text}%
  \label{sec:hint:hyphen-minus}}

In most cases, a "-" character inside \cs{text} will represent a minus. The
package will therefore replace it with \cs{textminus}. However, this may be
problematic if you want a hyphen, or if you are using "-" in a piece of \TeX{}
syntax. You can prevent this by using a second set of braces
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\electronvolt{\text{{e\kern -0.1em V}}}
\end{LaTeXdemo}
or by defining a protected command that will yield a hyphen on typesetting
\begin{LaTeXdemo}[code only]
  \usepackage{etoolbox}
  \newrobustcmd*\hyphenminus{-}
  \DeclareSIUnit\electronvolt{\text{e\kern \hyphenminus0.1em V}}
\end{LaTeXdemo}

\subsection{Expanding content in tables%
  \label{sec:hint:expanding}}

When processing tables, \pkg{siunitx} will expand anything stored inside a
macro, unless it is long or protected. \LaTeXe{} robust commands are also
detected and are not expanded (Table~\ref{tab:xmpl:macro}). Values which would
otherwise be expanded can be protected by wrapping them in a set of braces. As
\TeX{} itself will expand the first token in a table cell before \pkg{siunitx}
can act on it, using the \eTeX{} protected mechanism is the recommended course
of action to prevent expansion of macros in table cells. (As is shown in the
table, \TeX's expansion of \LaTeXe{} robust commands can lead to unexpected
results.)
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Values as macros in \texttt{S} columns.%
      \label{tab:xmpl:macro}}
    \newcommand*\myvaluea{1234}
    \newcommand\myvalueb{1234}
    \DeclareRobustCommand*\myvaluec{1234}
    \protected\def\myvalued{1234}
    \begin{tabular}{@{}S@{}}
    \toprule
      {Some Values} \\
    \midrule
       \myvaluea  8.8 \myvaluea  \\ % Both expanded
       \myvalueb  8.8 \myvalueb  \\ % First expanded by TeX
                                    % to numbers
       \myvaluec  8.8 \myvaluec  \\ % First expanded by TeX
                                    % but not to numbers!
       \myvalued  8.8 \myvalued  \\ % Neither expanded
      {\myvaluea\ 8.8 \myvaluea} \\ % Neither expanded
    \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

It is possible to use calculated values in tables. For this to work, the
calculation must take place before attempting to parse the number (the parser
cannot \enquote{know} all of the possible values inside an expression). This is
most conveniently handled using the \pkg{xfp} package, which is distributed as
part of the required support for \pkg{siunitx}. The general approach is
illustrated in Table~\ref{tab:xmpl:calc}
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Calculated values.%
      \label{tab:xmpl:calc}}
    \newcommand{\valuea}{66.7012}
    \newcommand{\valueb}{66.0212}
    \newcommand{\valuec}{64.9026}
    \begin{tabular}{
      @{}
      S[table-format = 2.4]
      S[table-format = 3.4]
      @{}
    }
      \toprule
        {Value} & {Doubled} \\
      \midrule
        \valuea & \fpeval{\valuea * 2} \\
        \valueb & \fpeval{\valueb * 2} \\
        \valuec & \fpeval{\valuec * 2} \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

A more sophisticated approach is to generate the rows themselves from a
database: this is illustrated in Section~\ref{sec:hint:datatool}.

\subsection{Using \pkg{siunitx} with \pkg{datatool}%
  \label{sec:hint:datatool}}

As illustrated in Table~\ref{tab:xmpl:calc}, \pkg{siunitx} can be used to
typeset data stored using \pkg{datatool}. For quickly displaying the contents
of tables, \pkg{datatool} offers the \cs{DTLshowtable} macro. This will only
work with \texttt{S} columns if number parsing is turned off
(Table~\ref{tab:xmpl:datatool}).
\begin{LaTeXdemo}[code and float]
  \DTLnewdb{moredata}
  \DTLnewrow{moredata}\DTLnewdbentry{moredata}{value}{ 6.7012}
  \DTLnewrow{moredata}\DTLnewdbentry{moredata}{value}{66.0212}
  \DTLnewrow{moredata}\DTLnewdbentry{moredata}{value}{64.902 }
  \begin{table}
    \caption{Displaying a \textsf{datatool} table.%
      \label{tab:xmpl:datatool}}
    \sisetup{parse-numbers= false, table-format = 2.4}
    \renewcommand*\dtlrealalign{S}
    \DTLdisplaydb{moredata}
  \end{table}
\end{LaTeXdemo}

The \pkg{datatool} package may also be used to create on-the-fly tables using
calculations. For example, the demonstration in Table~\ref{tab:xmpl:calc} may
be achieved without needing to write out each row, as shown in
Table~\ref{tab:xmpl:datatool-calc}. An extra column is used to contain the
calculations in this case.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Calculated values using \pkg{datatool}.%
      \label{tab:xmpl:datatool-calc}}
    \DTLnewdb{data}
    \DTLnewrow{data}\DTLnewdbentry{data}{value}{66.7012}
    \DTLnewrow{data}\DTLnewdbentry{data}{value}{66.0212}
    \DTLnewrow{data}\DTLnewdbentry{data}{value}{64.9026}
    \begin{tabular}{
      @{}
      S[table-format = 2.4]
      S[table-format = 3.4]
      @{}l
      @{}
    }
      \toprule
        {Value} & {Doubled} &
      \DTLforeach{data}{\myvalue=value}{%
        \DTLiffirstrow {\\ \midrule}{\\}%
        \myvalue &            % First column
        \fpeval{\myvalue * 2} % second column
        & }\\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\subsection{Using units in section headings and bookmarks%
  \label{sec:hint:bookmarks}}

The \pkg{siunitx} code is designed to work correctly with functions in
headings. They will print correctly in headings and in the table of contents.
As illustrated here, the standard behaviour is to ignore font changes. When the
\pkg{hyperref} package is loaded, the functions automatically \enquote{degrade
gracefully} to produce useful information in \acro{pdf} bookmarks. If you want
more control over the bookmark text, use the \cs{texorpdfstring} function from
\pkg{hyperref}, for example:
\begin{LaTeXdemo}[code only]
  \section{Some text
    \texorpdfstring
      {\unit{\joule\per\mole\per\kelvin}}
      {J mol-1 K-1}%
    }
\end{LaTeXdemo}

You may find it useful to load \pkg{hyperref} with the \opt{unicode} option,
as this will allow \unit{\ohm} to appear in bookmarks. Without the option,
the encoding used by \pkg{hyperref} does not support this symbol.

\subsection{A left-aligned column visually centred under a heading%
  \label{sec:hint:left-column}}

When you have a column of non-related numbers, the usual advice is to make
these left-aligned and then centre the resulting column under the heading. With
the \pkg{dcolumn} package, that would be done with something like
|D{x}{}{5.0}|. This is something of an abuse of the nature of a number, but can
also be done using \pkg{siunitx} (Table~\ref{tbl:xmpl:unrel}).
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Formatting unrelated numbers.%
      \label{tbl:xmpl:unrel}}
    \centering
    \begin{tabular}
      {
        @{}
        S[
          table-format = 5.0,
          parse-numbers = false,
          input-decimal-markers = x
        ]
        @{}
      }
      \toprule
      \multicolumn{1}{@{}c@{}}{Header} \\
      \midrule
      120   \\
      12.3  \\
      12340 \\
      12.02 \\
      123   \\
      1     \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\subsection{Regression tables%
  \label{sec:hint:regression}}

In some subject areas, it is common to present regression values or similar,
which feature an uncertainty value in parenthesis on the line below the main
value. As these are separate cells, they cannot be entered using \pkg{siunitx}
in one value. There are a couple of ways of formatting them using the package,
depending on whether the values also need to be parsed.

Where parsing is not required, the most straight-forward method is available:
provide a model format allowing space for an extra \enquote{digit} at each end,
which will then allow for the parenthesis. If a sign is applied to the number,
it may not be necessary to add a \enquote{digit} for the leading bracket. If
parsing is also required, this approach cannot be employed. Instead, the
parsing needs to be adjusted such that |(| and |)| are not treated as part of
the number, and \opt{table-align-text-before} is set to |false| such that these
will be placed next to the numerical part. These methods are illustrated in
Table~\ref{tab:regression}.

\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Regression tables%
      \label{tab:regression}
    }
    \begin{tabular}
      {
        @{}
        S[table-format = 2.4, parse-numbers = false]
        S[table-format = +1.4, parse-numbers = false]
        S[
          input-open-uncertainty  = ,
          input-close-uncertainty = ,
          minimum-decimal-digits  = 3, % (
          table-format            = +1.3),
          table-align-text-before = false
          ]
        @{}
      }
      \toprule
      {Header} &{Header} & {Header} \\
      \midrule
      1.234   & -1.234  & -1.23   \\
      (0.053) & (0.053) & (0.053) \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\subsection{Maximising performance%
  \label{sec:hint:performance}}

Both the number and unit parsers require significant effort in terms of \TeX{}
programming. For input that does not require such processing, the maximum
performance for \pkg{siunitx} can therefore be obtained by turning off both
systems:
\begin{LaTeXdemo}
  \qty{7.3}{\Hz} \\
  \qty[parse-units = false]{7.3}{\Hz} \\
  \qty[
    parse-numbers = false,
    parse-units   = false
  ]{7.3}{\Hz}
\end{LaTeXdemo}

For tables, any settings that can be given before the table are only parsed once,
whereas given in the optional argument to |S| they are read in every cell. As such,
you should favour
\begin{LaTeXdemo}[code only]
  \begin{table}
    \sisetup{...}
    \begin{tabular}{S}
    ...
\end{LaTeXdemo}
for common settings.

\subsection{Special considerations for the \cs{kWh} unit%
  \label{sec:hint:kWh}}

The standard settings provide the \cs{kWh} unit set up with no spacing between
the \enquote{\unit{\kilo\watt}} and the \enquote{\unit{\hour}} unit to give
\enquote{\unit{\kWh}}. However, this only applies when the unit is given on its
own: combinations will follow the normal rules
\begin{LaTeXdemo}
  \unit{\kWh} \\
  \unit{\kWh\per\metre}
\end{LaTeXdemo}
This is because the unit \cs{kWh} is defined so that it can still be varied by
altering \cs{kilo}, \cs{watt} and \cs{hour}, and so that the prefix can still
be turned into a number. However, some users may prefer to have a non-flexible
macro which never adds a space. This can be achieved by redefining \cs{kWh}
with \cs{DeclareSIUnit}, by added an alternative definition
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\kWh{kWh}
  \DeclareSIUnit\KWH{kWh}
\end{LaTeXdemo}
 or of course by using literal unit input.
\begin{LaTeXdemo}
  \unit{\KWH\per\metre}\\
  \unit{kWh.m^{-1}}
\end{LaTeXdemo}

Another point to notice is that the \cs{per} macro applies to the next unit,
and not an entire unit combination. Thus in
\begin{LaTeXdemo}
  \unit{\candela\per\kWh}
\end{LaTeXdemo}
\cs{per} applies to the watts but not to the hours. In this case, the units
need to be written out in full or the \opt{sticky-per} option should be used.
\begin{LaTeXdemo}
  \unit{\candela\per\kilo\watt\per\hour} \\
  \unit[sticky-per]{\candela\per\kWh}
\end{LaTeXdemo}

\subsection{Creating a column with numbers and units%
  \label{sec:hint:mixed}}

Usually, numbers in a table should be given with the units in the column
heading. However, there are cases where a series of data are best presented in
a table but have different units. There are two ways to do this
(Table~\ref{tab:xmpl:mixed}). The first is to place the units in the first
column of the table, which makes sense if there are several related items in
the table. The second method is to generate two columns, one for numbers and a
second for units, and then to format these to give the visual effect of a
single column. The later effect is most appropriate when only one set of
numbers are presented in a table. This method requires cell content is
collected, easiest to do using the \pkg{collcell} package.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Tables where numbers have different units.%
      \label{tab:xmpl:mixed}}
    \hspace{\fill}%
    \begin{tabular}
      {
        @{}
        >{$}l<{$}
        S[table-format = 3.3(1)]
        S[table-format = 3.3(1)]
        @{}
      }
      \toprule
        & {One} & {Two} \\
      \midrule
      a / \unit{\pm}         &  123.4(2) &   567.8(4) \\
      \beta / \unit{\degree} & 90.34(4)  & 104.45(5)  \\
      \mu / \unit{\per\mm}   &  0.532    &   0.894    \\
      \bottomrule
    \end{tabular}
    \hspace{\fill}%
    \begin{tabular}
      {
        @{}
        S[table-format=1.3]@{\,}
        >{\collectcell\unit}l<{\endcollectcell}
        @{}
      }
      \toprule
      \multicolumn{2}{@{}c}{Heading} \\
      \midrule
      1.234 & \metre   \\
      0.835 & \candela \\
      4.23  & \joule\per\mole \\
      \bottomrule
    \end{tabular}
    \hspace{\fill}%
  \end{table}
\end{LaTeXdemo}

\subsection{Tables with heading rows%
  \label{sec:hint:heading}}

A common format for tables is to make the heading row visually distinct using a
background color and bold text. If numbers appear in such a heading row within
an \texttt{S} column then getting the appearance right can be challenging. The
best approach is to make the \cs{bfseries} macro \enquote{robust} (as
demonstrated in Section~\ref{sec:hint:expanding}), then to use this macro to
make the heading cells bold. This approach is illustrated in
Table~\ref{tab:xmpl:headers}, along with the use of \cs{rowcolor} to provide a
background color. Some typical \LaTeX{} font arrangements use a wider
(extended) width for the bold font compared with the normal font. This will
disturb the alignment if the \opt{table-format} is not symmetrical. For these
cases, the \opt{table-model-setup} key may be used to apply the wider font to
the \enquote{model} used in ensuring alignment.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Header row in a table.%
      \label{tab:xmpl:headers}}
    \robustify\bfseries
    \centering
    \sisetup{text-series-to-math}
    \begin{tabular}
      {
        @{}
        S[table-format = 3.3]
        S[table-format = 3.6, table-model-setup = \bfseries]
        @{}
      }
      \rowcolor[gray]{0.9}
      \bfseries 123.456 & \bfseries 123.456789 \\
       23.45  &  23.45  \\
      123.4   & 123.4   \\
        3.456 &   3.456 \\
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\subsection{Associating a locale with a \pkg{babel} language%
  \label{sec:hint:babel}}

It is possible to instruct the \pkg{babel} package to switch to a particular
\pkg{siunitx} locale when changing language. This can be done using the
\pkg{babel} \cs{extras\meta{language}} system. For example, to associate the
\texttt{DE} locale with the \texttt{german} \pkg{babel} language, the
appropriate code would be
\begin{LaTeXdemo}[code only]
  \addto\extrasgerman{\sisetup{locale = DE}}
\end{LaTeXdemo}

\subsection{Symbolic \enquote{digits}%
  \label{sec:hint:symbols}}

In some cases you may want to use \enquote{digits} which do not fall within the
usual set |0123456789|. This can be done by setting the \opt{input-digits}
option, but bearing in mind that this will affect (prevent) for example
rounding.
\begin{LaTeXdemo}
  \sisetup{input-digits = 0123456789\pi}%
  \num{4\pi e-7}
\end{LaTeXdemo}
Each extra entry should be a single token, and should either have a definition
which is safe in both math and text mode, or should only be used when the
output mode is known. It may be necessary to make some tokens robust using
\pkg{etoolbox} for this to work, for example
\begin{LaTeXdemo}
  \robustify\dots
  \sisetup{input-digits = 0123456789\dots}%
  \qty{0,4066\dots}{\metre\squared}
\end{LaTeXdemo}

\subsection{Demonstrating prefixes%
  \label{sec:hint:prefixes}}

As \pkg{siunitx} contains data about the numerical values of unit prefixes,
you may wish to print this in an automated way. Prefixes cannot be given
on their own, but it is possible to create a \enquote{do nothing} unit.
\begin{LaTeXdemo}[code only]
  \DeclareSIUnit\noop{\relax}
\end{LaTeXdemo}
which can then be used to show just the prefix symbol.
\begin{LaTeXdemo}
  \unit{\yotta\noop}
\end{LaTeXdemo}
To show just the numerical value of a prefix, you will need to use \cs{qty}
and appropriate settings.
\begin{LaTeXdemo}
  \qty[prefix-mode = extract-exponent, print-unity-mantissa = false]%
    {1}{\yotta\noop}
\end{LaTeXdemo}
This may be conveniently wrapped up inside a document command, for example
\begin{LaTeXdemo}[code only]
  \NewDocumentCommand\prefixvalue{m}{%
    \qty[prefix-mode=extract-exponent,print-unity-mantissa=false]{1}{#1\noop}
  }
\end{LaTeXdemo}

\subsection{Creating a set of pre-defined units%
  \label{sec:hint:config}}

There are many units which sit outside of those defined in the (current)
\acro{SI} Brochure which are of use to many people. Most obvious are those
which have been detailed in previous editions of the Brochure, as described
in Section~\ref{sec:BIPM8}, but there are many others.

It is often convenient to have a pre-defined set of useful units available
without needing to copy the full set of definitions into each source file. At
the same time, it is important that such sources do show that they are using
units not defined by the core part of \pkg{siunitx}. The most straight-forward
way to achieve this is to create a separate file, for example
\texttt{siunitx-local-units.tex}, and place it in your local \TeX{} tree
(usually \verb|~/texmf/tex/latex/| on Linux, \verb|~/Library/texmf/tex/latex/|
on macOS or \verb|C:\Users\<name>\texmf\tex\latex| on Windows). This can then
be loaded in the preamble using
\begin{LaTeXdemo}[code only]
  % Load useful 'local' units
  \input{siunitx-local-units}
\end{LaTeXdemo}

\subsection{Overloading the standard interfaces%
  \label{sec:hint:overload}}

The interfaces in \pkg{siunitx} deliberately split up different types of
numerical input (ranges, products, etc.) for semantic reasons. This also means
that there is less chance of confusion in parsing input. However, some users
would prefer to overload the \cs{num} and \cs{qty} commands such that they
accept multiple different types of input. There is a code-level interface that
could be used for this, but most end-users will likely want to avoid this. A
possible approach that covers most of the common cases is.
\begin{LaTeXdemo}[code only]
  \ExplSyntaxOn
  \cs_gset_eq:NN \ifstrpresent \tl_if_in:nnTF
  \ExplSyntaxOff
  \NewCommandCopy\oldnum\num
  \NewCommandCopy\oldnumrange\numrange
  \RenewDocumentCommand\num{O{}m}{%
    \ifstrpresent{#2}{x}
      {\numproduct[#1]{#2}}
      {%
        \ifstrpresent{#2}{;}
          {\numlist[#1]{#2}}
          {%
            \ifstrpresent{#2}{:}
              {\numrange[#1]{#2}}
              {\oldnum[#1]{#2}}%
          }%
      }%
  }
  \RenewDocumentCommand\numrange{O{}>{\SplitList{:}}m}
    {\oldnumrange[#1]#2}
\end{LaTeXdemo}
for the \cs{num} command and
\begin{LaTeXdemo}[code only]
  \ExplSyntaxOn
  \cs_gset_eq:NN \ifstrpresent \tl_if_in:nnTF
  \ExplSyntaxOff
  \NewCommandCopy\oldqty\qty
  \NewCommandCopy\oldqtyrange\qtyrange
  \RenewDocumentCommand\qty{O{}mm}{%
    \ifstrpresent{#2}{x}
     {\qtyproduct[#1]{#2}{#3}}
     {%
       \ifstrpresent{#2}{;}
          {\qtylist[#1]{#2}{#3}}
          {%
            \ifstrpresent{#2}{:}
              {\qtyrange[#1]{#2}{#3}}
              {\oldqty[#1]{#2}{#3}}%
          }%
      }%
  }
  \RenewDocumentCommand\qtyrange{O{}>{\SplitList{:}}mm}
    {\oldqtyrange[#1]#2{#3}}
\end{LaTeXdemo}
for \cs{qty}.

\section{Using (\acro{SI}) units}

Consistent and logical units are a necessity for scientific work, and have
applicability everywhere. Historically, a number of systems have been used for
physical units. \acro{SI} units were introduced by the \foreign{Conférence
Générale des Poids et Mesures} (\acro{CGPM}) in 1960. \acro{SI} units are a
coherent system based on seven base units, from which all other units may be
derived.

At the same time, physical quantities with units are mathematical entities, and
as such way that they are typeset is important. In mathematics, changes of type
(such as using bold, italic, sans serif typeface and so on) convey information.
This means that rules exist not only for the type of units to be used under the
\acro{SI} system, but also the way they should appear in print. Advice on best
practice has been made available by the \emph{National Institute of Standards
and Technology} (\acro{NIST})~\cite{NIST}.

As befits an agreed international standard, the full rules are detailed. It is
not appropriate to reproduce these in totality here; instead, a useful summary
of the key points is provided. The full details are available from the
\foreign{Bureau International des Poids et Mesures}~\cite{BIPM}.

The \pkg{siunitx} package takes account of the information given here, so far
as is possible. Thus the package defaults follow the recommendations made for
typesetting numbers and units. Spacing and so forth is handled in such a way as
to make implementing the rules (relatively) easy.

\subsection{Units}

There are seven base \acro{SI} units, listed in Table~\ref{tab:unit:base}.%
\footnote{Some base units need others defined first; there is therefore a
required order of definition.} The base units have been chosen such that all
physical quantities can be expressed using an appropriative combination of
these units, needing no others and with no redundancy.

All other units within the \acro{SI} system are regarded as \enquote{derived}
from the seven base units. At the most basic, all other \acro{SI} units can be
expressed as combinations of the base units. However, many units (listed in
Table~\ref{tab:unit:derived}) have a special name and symbol. Most of these
units are simple combinations of one or more base units (raised to powers as
appropriate).

A series of \acro{SI} prefixes for decimal multiples and sub-multiples are
provided, and can be used as modifiers for any \acro{SI} unit (either base or
derived units) with the exception of the kilogram. The prefixes are listed in
Table~\ref{tab:unit:prefix}. No space should be used between a prefix and the
unit, and only a single prefix should be used. Even the degree Celsius can be
given a prefix, for example \qty{1}{\milli\degreeCelsius}.

It is important to note that the kilogram is the only \acro{SI} unit with a
prefix as part of its name and symbol. Only single prefix may be used, and so
in the case of the kilogram prefix names are used with the unit name
\enquote{gram} and the prefix symbols are used with the unit symbol
\unit{\gram}. For example $\qty{1e-6}{\kilo\gram} = \qty{1e-3}{\gram} =
\qty{1}{\milli\gram}$.

The application of \acro{SI} units is meant to provide a single set of units
which ensure consistency and clarity across all areas. However, other units are
common is many areas, and are not without merit. The units provided by
\pkg{siunitx} by default do not include any of these; only units which are part
of the \acro{SI} set or are accepted for use with \acro{SI} units are defined.
However, several other sets of units can be loaded as optional modules. The
binary prefixes and units (Table~\ref{tab:unit:binary}) are the most obvious
example. These are \emph{not} part of the \acro{SI} specifications, but the
prefix names are derived from those in Table~\ref{tab:unit:prefix}.

Other units are normally to be avoided where possible. \acro{SI} units should,
in the main, be preferred due to the advantages of clear definition and
self-consistency this brings. However, there will probably always be a place
for specialist or non-standard units. This is particularly true of units
derived from basic physical constants.

There are also many areas where non-standard units are used so commonly that to
do otherwise is difficult or impossible. For example, most synthetic chemists
measure the pressure inside vacuum apparatus in \unit{\millimetremercury},
partly because the most common gauge for the task still uses a column of
mercury metal. For these reasons, \pkg{siunitx} allows definition of such
units.

\subsection{Mathematical meaning}

As explained earlier, a quantity combination is a single mathematical entity.
This has implications for how both the number and the unit should be printed.
Firstly, the two parts should not be separated: a quantity is a product of the
number and the unit. With the exception of the symbols for plane angles
(\unit{\degree}, \unit{\arcminute} and \unit{\arcsecond}), the \acro{BIPM}
specifies either a space or half-height (centred) dot should be
used~\cite{BIPM}.
\begin{LaTeXdemo}
  A space for \qty{10}{\percent}\\
  and also for \qty{100}{\degreeCelsius}\\
  but not for \ang{1.23}.
\end{LaTeXdemo}

The mathematical meaning of units also means that the shape, weight and family
are important. Units are supposed to be typeset in an upright, medium weight
font. Italic, bold and sans serif are all used mathematically to convey other
meanings. (In an all sanserif document, using sans serif for units is
reasonable.) The \pkg{siunitx} package defaults again follow this convention:
any local settings are ignored, and uses the current upright math font.
However, there are occasions where this may not be the most desirable
behaviour. A classic example would be in an all-bold section heading. As the
surrounding text is bold, some people feel that any units should follow this.
\begin{LaTeXdemo}
  Units should \textbf{not be bold: \qty{54}{\farad}}\\
  \textbf{But perhaps in a running block,\\
  it might look better:
    \qty[text-series-to-math]{54}{\farad}}
\end{LaTeXdemo}

Symbols for units formed from other units by multiplication are indicated by
means of either a half-height (that is, centred) dot or a (thin) space.
\begin{LaTeXdemo}
  $\unit{\metre\second} = \text{metre second}$ \\
  $\unit{\milli\second} = \text{millisecond}$ \\
  \sisetup{inter-unit-product = \ensuremath { { } \cdot { } } }
  $\unit{\metre\second} = \text{metre second}$ \\
  $\unit{\milli\second} = \text{millisecond}$
\end{LaTeXdemo}
There are some circumstances under which it is common practice to omit any
spaces. The classic example is \unit{\kWh}, where
\enquote{\unit[inter-unit-product = \,]{\kWh}} does not add any useful
information. If using such a unit repeatedly, users of \pkg{siunitx} are
advised to create a custom unit to ensure consistency. It is important to note
that while this is common practice, it is \emph{not} allowed by the
\acro{BIPM}~\cite{BIPM}.

Symbols for units formed from other units by division are indicated by means of
a virgule (oblique stroke, slash, "/"), a horizontal line, or negative
exponents.\footnote{Notice that a virgule and a solidus are not the same
symbol.} However, to avoid ambiguity, the virgule must not be repeated on the
same line unless parentheses are used. This is ensured when using named unit
macros in \pkg{siunitx}, which will \enquote{trap} repeated division and format
it correctly. In complicated cases, negative exponents are to be preferred over
other formats.
\begin{LaTeXdemo}
  \unit{\joule\per\mole\per\kelvin} \\
  \unit[per-mode = fraction]{\joule\per\mole\per\kelvin} \\
  \unit[per-mode = symbol]{\joule\per\mole\per\kelvin}
\end{LaTeXdemo}

Products and errors should show what unit applies to each number given. Thus
\qtyproduct[product-units = bracket]{2x3}{\metre} is an ordered set of lengths
of a geometric area, whereas \qtyproduct[product-units = single]{2x3}{\metre}
is a length (and equal to \qty{6}{\metre}). Thus, $\times$ is not a product but
is a mathematical operator; in the same way, a $2 \times 3$ matrix is not a $6$
matrix! In some areas, areas and volumes are given with separated units but a
unit raised to the appropriate power: \qtyproduct[product-units = power]{2 x
3}{\metre}. Although this does display the correct overall units, it is
potentially-confusing and is not encouraged.

Care must be taken when writing ranges of numbers. For purely numerical values,
it is common to use an en-dash to show a range, for example \enquote{see pages
1--5}. On the other hand, physical quantities could be misinterpret as negative
values if written in this way. As the quantity is a single mathematical entity,
writing the values with an en-dash followed by a single unit is also incorrect.
As a result, using the word \enquote{to} is strongly recommended.
\begin{LaTeXdemo}
  \qtyrange{1}{5}{\metre} long.
\end{LaTeXdemo}

\subsection{Graphs and tables}

In graphs and tables, repetition of the units following each entry or axis mark
is confusing and repetitive. It is therefore best to place the unit in the
label part of the information. Placing the unit in square brackets is common
but mathematically poor.\footnote{For example, for an acceleration \(a\), the
expression $[a]$ is the dimensions of $a$, \foreign{i.e.}~length per time
squared in this case.} Much better is to show division of all quantities by the
unit, which leaves the entries as unitless ratios. This is illustrated in
Table~\ref{tab:xmpl:unitless} and Figure~\ref{fig:xmpl:unitless}.
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{An example of table labelling.%
      \label{tab:xmpl:unitless}}
    \sisetup{
      table-number-alignment = center,
      table-format = 1.4
    }
    \begin{tabular}{@{}cS@{}}
      \toprule
        Entry & {Length/\unit{\metre}} \\
      \midrule
        1 & 1.1234 \\
        2 & 1.1425 \\
        3 & 1.7578 \\
        4 & 1.9560 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}
\begin{LaTeXdemo}[code and float]
  \begin{figure}
    \begin{tikzpicture}
      \begin{axis}[
        xlabel = $t/\unit{\second}$,
        xmax   = 6,
        xmin   = 0,
        ylabel = $d/\unit{\metre}$,
        ymin   = 0
      ]
        \addplot[smooth,mark=*]
          plot coordinates {
            (0,0)
            (1,5)
            (2,8)
            (3,9)
            (4,8)
            (5,5)
            (6,0)
           };
      \end{axis}
    \end{tikzpicture}
    \caption{An example of graph labelling.%
      \label{fig:xmpl:unitless}}
  \end{figure}
\end{LaTeXdemo}

 In most cases, adding exponent values in the body of a table is
 less desirable than adding a fixed exponent to column headers.  An
 example is shown in Table~\ref{tab:good}.  The use of \cs{multicolumn} is
 needed here due to the \enquote{\texttt{<}}; without \cs{multicolumn},
 the titles are followed by \enquote{\unit{\kilo\gram}}!
\begin{LaTeXdemo}[code and float]
  \begin{table}
    \caption{Bad and good columns.%
      \label{tab:good}}
    \sisetup{table-number-alignment = center}
    \begin{tabular}{
      @{}
      c
      S[table-format = 1.3e1]
      @{\,\unit{\kilogram}}
      S[table-format = 2.2]
      @{}
    }
      \toprule
        Entry & \multicolumn{1}{c}{Mass} &
          {Mass/\qty[print-unity-mantissa = false]{e3}{\kilogram}} \\
      \midrule
        1 & 4.56e3  & 4.56 \\
        2 & 2.40e3  & 2.40 \\
        3 & 1.345e4 & 13.45 \\
        4 & 4.5e2   & 0.45 \\
      \bottomrule
    \end{tabular}
  \end{table}
\end{LaTeXdemo}

\section{Installation}

For most users, there will be no need to explicitly install \pkg{siunitx}:
it is available from the package management system in current \TeX{} Live
and MiK\TeX{} systems.

For manual installation, the package is available from
\href{http://ctan.org/pkg/siunitx}{\acro{CTAN}}. As well as the raw source
files, \acro{CTAN} hold the package as a pre-extracted zip file,
\file{siunitx.tds.zip}. The later is most convenient for most users: simply
unzip this in your local \path{texmf} directory.

The package requires \pkg{expl3} support as provided in the \pkg{l3kernel} and
bundle, which is required by the \LaTeX{} kernel. As such, you are very
unlikely to need to update this manually.

\section{Thanks}

Many users have provided feedback, bug reports and ideas for new features for
\pkg{siunitx}: thanks to all of them. Particular thanks to Stefan Pinnow, who
has taken the lead role as beta tester for \pkg{siunitx}, finding incorrect
output, bad documentation and the odd spelling mistake in the documentation.
Thanks also to Enrico Gregorio for encouraging me to complete a fully
\pkg{expl3}-compliant version of the package. Thanks also to Danie Els and
Marcel Heldoorn for the \pkg{SIstyle} and \pkg{SIunits} packages, respectively,
which provided the starting point for the development of \pkg{siunitx}.

\section{Making suggestions and reporting bugs}

Feedback on \pkg{siunitx} is always welcome, either to make suggestions or to
report problems. When sending feedback, it is always useful if a small example
file is included, showing the bug being reported or illustrating the desired
output. It is helpful if a \enquote{reference rendering} is included, showing
what the output should look like. A typical example file might read
\begin{verbatim}
  \listfiles
  % Use the article class unless the problem is class-dependent
  \documentclass{article}
  \usepackage{siunitx}
  % Other packages loaded as required
  \begin{document}
  Reference output: $1.23\,\mathrm{m}$

  \textsf{siunitx} output: \qty{1.23}{\metre}
  \end{document}
\end{verbatim}
As illustrated, it is usually best to use the \cls{article} class and to only
load packages which are needed to show the issue. It is also useful to include
a copy of the log file generate by \LaTeX{} when reporting a bug (as the
versions of packages can be important to solving the issue).

Feedback can be sent in a range of ways. The development code and issue tracker
are hosted on GitHub: \url{https://github.com/josephwright/siunitx/}. Issues
opened there are visible to other users and makes sure that they cannot be
forgotten.

\end{documentation}

\begin{thebibliography}{10}
  \bibitem{BIPM}
    \emph{The International System of Units (\acro{SI})},
    \url{https://www.bipm.org/en/measurement-units/}.
  \bibitem{NIST}
    \emph{International System of Units from \acro{NIST}},
    \url{http://physics.nist.gov/cuu/Units/index.html}.
\end{thebibliography}

\PrintIndex

\end{document}