I am writing a book in Rmarkdown using the tufte package; which uses tufte-latex for pdf output. My current issue is that I have not found a way to have a table extend wider than the normal margin of the tufte-latex output.

Things I have tried:

  • Wrapping in the table* environment (suggested here); but works only for tables that are less than 1 page long
  • Wrapping in the fullwidth environment; but this has no effect on longtable (see here)
  • Using my newgeometry approach, which pushes the table onto a new page, which is not desirable
  • \setlength\LTleft and \setlength\LTright (here); I can set one or the other (e.g. move the table to the right margin), but setting both doesn't alter the width. I think this is because the column sizes are fixed by Pandoc?

Things I have read but don't understand:

  • Redefining the longtable environment here looks promising, but I don't entirely understand the answer

Things I can do:

  • I can use Pandoc filters to wrap latex output in additional commands or environment wrappers. e.g. I have a filter that identifies a wide table and places it in a \landscape layout (although this it does not solve the column-overlap issue).
  • I can change the latex template the Pandoc uses, so I can call additional options/use packages

Key restrictions are that:

  • Tables compile cleanly to both html and latex; i.e. it can take a markdown table as an input, and does not require explicit specification of column width
  • It can handle tables longer than one page

Here is an example table/document, generated by pandoc.


% ams

\usepackage{fixltx2e} % provides \textsubscript
\ifnum 0\ifxetex 1\fi\ifluatex 1\fi=0 % if pdftex
\else % if luatex or xelatex


% graphix

% booktabs

% url

% hyperref

% units.


% citations

% pandoc syntax highlighting

% table with pandoc
\usepackage{calc} % for calculating minipage widths
% Correct order of tables after \paragraph or \subparagraph
% Allow footnotes in longtable head/foot

% multiplecol

% strikeout

% morefloats

% tightlist macro required by pandoc >= 1.14

% title / author / date
\title{A Minimal Book Example}
\author{Author Name}






  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.04}}
  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.20}}
  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.20}}
  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.24}}
  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.18}}
  >{\raggedright\arraybackslash}p{(\columnwidth - 10\tabcolsep) * \real{0.13}}@{}}
Properties & Dobutamine & Isoprenaline & Ephedrine & Metaraminol & Phenylephrine \\
Uses & Stress testing, increasing CO & Severe bradycardia & ↑ SVR without ↓ in HR & ↑ SVR & ↑ SVR \\
Dosing & 5-15µg/kg/min & Infusion from 0.5-10µg/min & 3-6mg bolus & Bolus 0.5-2mg & Bolus start at 50-100mcg \\
Route & IV & IV & IV & IV & IV/IM/SC \\
Presentation & Racemic mixture of 250mg dobutamine in 20ml water & Clear solution at 1mg/ml & Clear, colourless solution in 30mg/ml ampoule & Clear, colourless solution in ampoule at 10mg/ml, typically reconstituted to 0.5mg/ml & Clear, colourless solution at 100mcg/ml \\
Absorption & IV & IV & IV or IM & IV & IV \\
Metabolism & t1/2 2-5 min. COMT to inactive metabolites. & Hepatic by COMT & Hepatic (not metabolised by MAO and COMT), giving a longer (10-60 minute) duration of action and a t1/2β of 3-6 hours & Some uptake into adrenergic nerve endings & Hepatic by MAO \\
Elimination & Urinary excretion of unchanged drug and metabolites & 50\% unchanged in urine & Renal of metabolites, t1/2β 2-3 hours & & \\
Mechanism of action & β1\textgreater\textgreater β2, D2 & β1\textgreater β2 & ↑ NA release (indirect α1) and direct α and β agonism & Direct and indirect (via ↑ NA release) α1 agonism & Direct α1 \\
Respiratory & Bronchodilation & Potent bronchodilation & Bronchodilation & & \\
CVS & ↑ HR, CO, contractility, and automaticity. Β2 effects may ↓ SVR and BP, particularly if ↓ preload. & ↑ HR and ↑ CO, modest ↑ inotropy. ↓ SVR due to β2 effects but BP usually unchanged due to ↑ CO. & Direct and indirect (via NA release) ↑ in HR, BP, and CO. Arrhythmogenic. & ↑ SVR/PVR, reflex bradycardia. Indirect ↑ in coronary flow. & ↑ SVR and BP, potential reflex bradycardia. Not arrhythmogenic. \\
CNS & Tremor & & ↑ MAC, mydriasis. & & \\
MSK & & & & & \\
Renal & ↑ RBF and ↑ urinary output with no improvement in renal function & & ↓ RBF & ↓ RBF & ↓ RBF \\
Metabolic & & & & & \\
GU & & & & ↓ Uterine blood flow & ↓ Uterine blood flow \\
GIT & & & & & \\


1 Answer 1


All that is required is to expand the columns of the table to use the \marginparwidth and the \marginparsep.


\setlength{\tabcolsep}{6pt} % added to fit in the textwidth<<<<<<<<<<<<<<  
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.25}}
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.25}}
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.25}}
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.3}}
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.3}}
    >{\raggedright\arraybackslash}p{(\textwidth - 10\tabcolsep) * \real{0.3}}@{}}

The second page using \usepackage{showframe} to display the margins.


  • Hi Simon, thanks for your answer. Unfortunately I cannot change the numeric values of \real; they are generated by Pandoc when it compiles. Your answer has had me thinking though and I am wondering if I can redefine \real to scale by a fixed value (e.g. 1.6, which would fit the fullwidth of the document), and then un-redefine it after the table. I've had a bit of a play but my latex knowledge isn't good enough to have it working or see if its worth pursuing: \let\oldreal\real \renewcommand{\real}[2]{1.6 * \oldreal{#1}{#2}}
    – cjdbarlow
    Dec 22, 2021 at 23:11
  • @cjdbarlow That would not work. The first column is to narrow. Why? Try to force Pandoc or R so that all \real have the values you want. Dec 23, 2021 at 13:02
  • This put me on the right track. I wrote Pandoc filter that broke up tables based on their relative number of columns and rows; and made tables that were less than one page long fullwidth using \table* and landscaped everything bigger than that.
    – cjdbarlow
    Dec 29, 2021 at 8:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .