# Why exactly is \linespread{1.05} appropriate for Palatino?

Section 1.4 of l2tabu, Changing inter-line space using \baselinestretch, contains a remark about the Palatino font:

[I]f you only want to use fonts other than Computer Modern you may use \linespread{<factor>}. For example, when using Palatino \linespread{1.05} would be appropriate.

A factor of 1.05 for Palatino is also recommended in, e.g., this posting at comp.text.tex and (by implication) in section 2.1 Fundamentals of Page Layout of the KOMA-Script manual.

A possible reasoning is that the increased \linespread compensates for Palatino's larger x-height (height of lowercase letters without ascenders) and hereby avoids "cramped" text lines.

Palatino's x-height is about 9% larger than that of Computer Modern (4.69pt v. 4.31pt for a fontsize of 10pt, see MWE below). That is, the "recommended" Palatino \linespread factor of +5% is about half as large as the difference in x-height.

I'd like to know the following:

• Who exactly came up with \linespread{1.05} for Palatino? The authors of l2tabu or someone else?

• Is the factor of +5% just a rule of thumb, or is it the result of some (unknown to me) established typographic formula?

\documentclass{article}

\begin{document}

Computer Modern's x-height: \the\fontdimen5\font

\fontfamily{ppl}\selectfont

Palatino's x-height: \the\fontdimen5\font

\end{document}

-
I don't remember where I've seen 1.05 for the first time. I don't think there's a "rule" either. For example, if the language is German, even 1.08 might be a choice, because of the abundance of uppercase letters. –  egreg Oct 9 '11 at 21:27
The 1.05 probably comes from whoever made the first package for using Palatino with LaTeX (the now obsolete palatino package), i.e. Walter Schmidt. –  Martin Schröder Oct 9 '11 at 22:31
There is no (mathematical) rule for this. The general rule is that the wider the \textwidth, the larger the \baselineskip. Also the larger the x-height, the larger the \baselineskip. The combination of line width and leading should make sure 1) the reader can find the next line easily 2) the reader can follow the current line without distracted by following lines. The relations between x-height and leading is more arguably. Some types are designed to be "economically", they look not too bad if you set them in long lines and relatively small leading, while their x-height is also large. –  Yan Zhou Oct 9 '11 at 22:45
@Mico It's not just a question of color; one can choose Palatino because it's darker than CM. The \linespread method is too coarse when dealing with size changing problems: it's not automatic that, with Palatino set 10/12.6 for the main text, it's right at 9/11.6 for quotations (\small) or 8/10.5 for footnotes. But that would require writing a very detailed specification based on size10.clo. –  egreg Oct 10 '11 at 10:10
@Mico I think "readability" does not "outweigh" any concerns concerns. I believe that the considerations of colors actually should serves the purpose of readability. After all, all typographical considerations should serve the purpose of the text. If the texts are meant to be read, then the color, the font, all things should serve this purpose. If the texts are meant to be displayed, then everything serve this purpose and readability is not so important anymore. The purpose of the texts is the core to everything. Typography considerations are the tool not the goal. –  Yan Zhou Oct 10 '11 at 11:56

Some preliminary remarks. I think we can all agree that there can be no universally agreed-upon optimal amount of "typographic color" or "overall grayness" of a page of text. What some might consider to be "good color" may strike others as unpleasantly light, or too dark. Nevertheless, for the sake of the argument, let's assume that (i) there is an optimal amount of color and (ii) the settings for 10pt Computer Modern set "single-spaced" achieve this optimal amount. (By the way, this does not imply that the optimal distance between consecutive lines is 10pt for text set in CM. The topic of what exactly constitutes "single-spacing" is a matter for a different discussion.)

The question then becomes: how does one achieve this amount of color when switching to a different (text) font, such as Palatino? Palatino and CM obviously differ in many respects. Not only are their x-heights different, but so are their cap-heights and ascender heights as well as the average stroke width (to name just a couple of additional factors). The upshot is that if text that consists of a group of paragraphs is set both in Palatino and in CM with the same nominal point size (say, 10pt) and the same interline distance (say, 12pt), the paragraphs and pages set in Palatino will look noticeably darker. I think most will agree that the Palatino paragraphs will be "too dark" -- again, assuming that the text set in CM have "correct color".

• Who exactly came up with \linespread{1.05} for Palatino? The authors of l2tabu or someone else?
• Is the factor of +5% just a rule of thumb, or is it the result of some (unknown to me) established typographic formula?

Regarding your first question: I have no idea who was the first to come up with this recommendation. Most likely, though, the need to increase the "leading" when typesetting text in Palatino was "discovered" immediately after people started mixing/matching Palatino and CM.

The +5% recommendation you noted in the l2tabu document can be nothing more than a rule of thumb. My understanding is that it should be understood as a "lower bound" on the required line spread adjustment, rather than as a single number. The code below generates two half-pages of text; the upper half is in Latin Modern, the lower is in Palatino nova with a linespread factor of 1.05. (Both texts are set with nominal font sizes of 10pt, and the text itself is the usual lipsum stuff...) To my eyes at least, the text set in Palatino still has more color than the CM reference text despite the 5% adjustment. Playing with the line spacing setting, I'd say that ca. 7% achieves equal color.

% !TEX TS-program = xelatex
\documentclass[letterpaper,10pt]{article}
\frenchspacing\pagestyle{empty}
\usepackage{fontspec}
\usepackage[margin=1in]{geometry}
\usepackage{setspace,lipsum}

\begin{document}
\setmainfont{Latin Modern Roman}
\emph{Latin Modern}

\lipsum[1-3]

\vspace{1in}
\setmainfont{Palatino nova Regular}
\setstretch{1.05}

\lipsum[1-3]
\end{document}


-
Add some inline math! IMHO this is one of the main reasons that one needs to adjust the linespread for palatino. –  Yiannis Lazarides Oct 10 '11 at 7:47
@YiannisLazarides: I see your comment as advocating an increase in linespread even beyond 1.07 when using Palatino, right? –  Mico Oct 10 '11 at 10:28
not really more caveat emptor:) –  Yiannis Lazarides Oct 10 '11 at 12:00

Quoting Martin Schröder's comment:

The 1.05 probably comes from whoever made the first package for using Palatino with LaTeX (the now obsolete palatino package), i.e. Walter Schmidt.

I wasn't able to locate a version of palatino.sty that contains \linespread{1.05}, but Walter Schmidt indeed suggested this factor in a posting at de.comp.text.tex in March 2002. Here's my translation of the relevant parts:

The reasonable value depends on the width of the text line. \linespread{1.05} is a sound reference point. Try smaller and larger values, up to about 1.08, and you'll see what is best for readability.

\linespread{1.05} for Palatino is also recommended in the German version of lshort, co-authored by Walter Schmidt (but not in the adapted English version).

-
Interesting that the English and German versions of "lshort" use different values. IMHO is on purpose as one has more accented characters in German. Also I think for mathematical texts using a lot of inline math one should use 1.10. –  Yiannis Lazarides Oct 11 '11 at 23:24