# consistently increased line spacing when subscripts occur in paragraph

The question is about consistent vertical spacing in a paragraph of text that includes some math using subscripts.

Concretely, in the example I feel that there is not enough vertical space between line 1 and line 2, due to the fact there is both a subscript in the math on line 1 and some math as well on line 2.

In general, I would not want to fix this locally, but rather have a bit more vertical space consistently throughout the paragraph (whenever the situation occurs at least once in the pargraph).

Since I have a quite a lot of paragraphs where this situation occurs, it would be useful if I did not have to request that for every single paragraph, but if this could be a default.

\documentclass[draft,11pt]{article}
\sloppy

\usepackage{amssymb}

\begin{document}

\newcommand{\diam}[1]{\ensuremath{\langle #1 \rangle}}
\newcommand{\model}{\mathbb{M}}
\newcommand{\N}{\ensuremath{\mathsf{N}}}

Assume $\model|_{\N(w_0)\cap\ldots\cap\N(w_n)},w\models \diam{a}\psi$.
It follows that there is a state $v\in \N(w_0)\cap\ldots\cap\N(w_n)$
with $w R_a v$ (1) and $\model|_{\N(w_0)\cap\ldots\cap\N(w_n)},v\models \psi$ (2). Hence by IH  $\model| _{\N(w_0)\cap\ldots\cap\N(w_n)},v\models \psi$ (3). From (1) and the fact that $\model'$ is an $A$-generated
submodel of $\model$ we have $w R'_a v$, hence by (3) $\model'| _{\N(w_0) \cap\ldots\cap\N(w_n)},w\models \diam{a}\psi$. The other direction is
trivial.

\end{document}

• You could change \linespread, but if you do so, I would advise you to do it (for consistencyś sake) for all your document, in which case, the setspace package will be useful. – Gonzalo Medina Jan 23 '14 at 17:20
• But in all other cases I would not want the linespread to be bigger than what it is by default. So maybe I should accept to have lines a bit too close whenever I have subscripts. – sunless Jan 23 '14 at 17:27
• alternative is to change fontdimen so the subscripts are not lowered so much. tex.stackexchange.com/questions/88991/… – David Carlisle Jan 23 '14 at 17:35

In your example the baselines are at constant distance, as testified by this picture, where the horizontal rules are drawn independently of the text at \baselineskip distance from each other:

If your text has many subscripts, it can be a good idea to increase the leading; this is not bad per se, it is only if the increase is too big. Here are examples of the same text at normal leading, after a 5% increase and after a 10% increase. You can experiment with something in between; setting text at 11/15 (that is with \linespread{1.1}) is not a serious sin: typographic decisions depend on the nature of the text. Of course, the leading should be the same across the document, so \linespread{...} (without \selectfont) should go in the preamble.

Here's the code for producing the images

\documentclass[draft,11pt]{article}
\usepackage{amssymb}

\newcommand{\diam}[1]{\ensuremath{\langle #1 \rangle}}
\newcommand{\model}{\mathbb{M}}
\newcommand{\N}{\ensuremath{\mathsf{N}}}

\newcommand{\ruler}{%
\leavevmode
\llap{%
\smash{%
\raise\baselineskip\vtop to 5\baselineskip{
\vrule height\baselineskip width 0.1pt
\vrule width \textwidth height 0.1pt}\vfill
}%
}\kern\parindent
}%
}

\begin{document}

\newcommand{\sampletext}{%
Assume $\model|_{\N(w_0)\cap\ldots\cap\N(w_n)},w\models \diam{a}\psi$.
It follows that there is a state $v\in \N(w_0)\cap\ldots\cap\N(w_n)$
with $w R_a v$ (1) and $\model|_{\N(w_0)\cap\ldots\cap\N(w_n)},v\models \psi$ (2). Hence by IH  $\model| _{\N(w_0)\cap\ldots\cap\N(w_n)},v\models \psi$ (3). From (1) and the fact that $\model'$ is an $A$-generated
submodel of $\model$ we have $w R'_a v$, hence by (3) $\model'| _{\N(w_0) \cap\ldots\cap\N(w_n)},w\models \diam{a}\psi$. The other direction is
trivial.}

\noindent\textsf{The text has equally spaced baselines}\par
\medskip

\ruler\sampletext

\medskip
\noindent\textsf{The same text without a ruler}\par
\medskip

\sampletext

\medskip
\noindent\textsf{With slightly taller baseline skip, increased 5\%}\par
\medskip