[Tex/LaTex] Why is \topfraction larger than \bottomfraction

floats

Somebody I worked with told me that he does not like to have floats “here” ([h]) but only at “top”, “bottom”, or “page” ([tbp]). After using this for my own documents for a while, I quite like it; the “here” floats often are stumbing blocks for the reading flow.

The thing with figures in the “top” placement is that their caption (below the image) and the body text looks very alike, I have the same font and the scrreprt caption by default indents additional lines. I thought that it might be a good idea to restrict float placement to bp. Then the figure would not be a stumbling block within the text but the reader can also continue reading the body text at the beginning of the new page instead of finding the end of the caption first.

Noticing that all slightly larger floats end up on separate pages now, I found that \topfraction is way larger than \bottomfraction. It seems the authors of those default values have had some reason do chose them like this. Usually the defaults lead to visually pleasing results, so I wonder what the background is here.

Why are large floats at the bottom of a page discouraged by default?
Would I do the reader a favor by setting \bottomfraction to the value \topfraction has and using [bp] as a float placement?

Best Answer

Either the "turn of a page" or the "gutter" provides a physical break between the text and the float when they are at the top of the page. In my opinion, this prepares the reader better for the possibility of the discontinuity being interrupted by something like a float since the break could happen mid-paragraph.

Placing a float at the [b]ottom of the page may leave the reader jumping to the caption as part of the textual flow if there is no visual distinction between the paragraph and caption text (something like Figure X.Y: in bold or a different font/colour for the caption)^*.

This might be a (somewhat speculative) argument for discouraging floats at the bottom of a page in general. Setting \bottomfraction to be the same as \topfraction may also be in favour of placing content at the [b]ottom, may have little effect if precedence is given to placing floats at the [t]op. But then again, this depends heavily on the document layout.

_{^*The same goes for [h]ere (which in some ways are similar to [b]ottom) unless there is a visual queue for the break, like a paragraph separation.}

Related Solutions

[Tex/LaTex] Avoid showing all figures at the end of the text by setting \topfraction to a value larger than 1.0

The statements made in those tips are only partially correct and in my opinion should not be taken too seriously.

Correct is:

if a figure is larger than a full page then LaTeX changes the size of the figure to be \textheight (or rather make it appear so to the algorithm, the actual size is not changed) and gives a warning about this. So it will not prevent the float from fitting on a float page.
if floatpagefraction is larger than \topfraction then indeed a single float that has size between the two values can't be placed neither in the top area nor will it fit onto a float page without another float coming along (or a \clearpage). So in documents that have very few floats it might pose an issue. On the other hand, for many document scenarios such a setting can be quite sensible to avoid half empty float pages, especially in cases where the typical float sizes are small.
setting \topfraction to something larger than 1 (as suggested in the title of the question) does not make sense. In fact it doesn't even make sense to put it larger than (1-\textpagefraction). Otherwise you have inconsistent settings and that may potentially result in strange reactions from the algorithm.

A general comment here is that there aren't settings that can work well in all scenarios. The default settings used in the standard LaTeX classes (article, report, book) is trying to get the floats out reasonably fast at the cost of allowing half-empty float pages. For good layout that is rather loose, so adjusting it is often called for. In a final document one then may have to overwrite those setting on individual floats by using !-notation so that a float can move into a certain area even if the settings normally would prevent it.

As David said: a more extensive explanation on how the algorithm works is given in How to influence the position of float environments like figure and table in LaTeX?

[Tex/LaTex] Floats in the middle of paragraphs

This modifies LaTeX's top float positioning so that it puts the top floats in the centre of the page, or at least as close to the centre as possible, if the text flow has a an unbreakable box or math display at the centre the vsplit will not split it at exactly the centre. It's not extensively tested but it works on this page:-). It applies to all top floats in the document.

enter image description here

\documentclass{article}
\usepackage{kantlipsum}


\makeatletter

\def \@cflt{%
    \let \@elt \@comflelt
    \setbox\@tempboxa \vbox{}%
    \@toplist
\setbox\z@\vsplit\@outputbox to 0.5\ht\@outputbox
    \setbox\@outputbox \vbox{%
                             \boxmaxdepth \maxdepth
                             \unvbox\z@
                             \vskip .5\textfloatsep
                             \unvbox\@tempboxa
                             \vskip -\floatsep
                             \topfigrule
                             \vskip .5\textfloatsep
                             \unvbox\@outputbox
                             }%
    \let\@elt\relax
    \xdef\@freelist{\@freelist\@toplist}%
    \global\let\@toplist\@empty
}

\makeatother

\begin{document}
\kant[1-5]

\begin{figure}[t]

\rule{3cm}{6cm}

\caption{a rectangle}
\end{figure}

\kant[2-7]

\end{document}

Best Answer

Related Solutions

[Tex/LaTex] Avoid showing all figures at the end of the text by setting \topfraction to a value larger than 1.0

[Tex/LaTex] Floats in the middle of paragraphs

Related Question