TeX - LaTeX Stack Exchange is a question and answer site for users of TeX, LaTeX, ConTeXt, and related typesetting systems. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Well, let me explain. I tried to select a part of a text on a PDF file to copy and paste. I noticed that accented letters disappear when selected, as the image below shows.

enter image description here

So, I decided to see the properties of the PDF file to check the used fonts and I found this:

enter image description here

The PDF was created with pdflatex on TeXLive 2011 and the MWE is:



The PDF viewer is evince 3.2.1.

The use of latin1 instead of utf8 is that utf8 does not allow me to insert accented characters, like õ. It produces the error

inputenc: Keyboard character used is undefined(inputenc) in inputencoding `utf8'. dimensõ

Loading fontenc

So I'm using latin1 and also I loaded the fontenc package with T1 option. The output is OK and now I can select the accented letters, as the image shows:

enter image description here

Then I decided to check PDF properties again and here is the surprising fact: the embedded font is another font! Why??

enter image description here

share|improve this question
cmr10.pfb doesn't cover the accented glyphs, while sfrm1095 does and is the embedded font when T1 is chosen. – egreg Mar 17 '13 at 13:51
Could I have some problem with fonts if I share my TeX or PDF files with others? Also, why should I use the fontenc package if the result is the same (except by the selection problem)? – Sigur Mar 17 '13 at 13:54
If they have the CM-Super fonts installed on their machine, the result will be the same. You might want to try the Latin Modern fonts (\usepackage{lmodern}). – egreg Mar 17 '13 at 13:56
Amazing! With lmodern I don't need inputenc neither fontenc! Is this true? – Sigur Mar 17 '13 at 14:00
No. If you want to use non-ascii-chars in your input you should use inputenc. And you should use fontenc with T1-encoding - for copy & paste and other reasons. (Well to be more precise you could also use an encoding like LY1.) – Ulrike Fischer Mar 17 '13 at 14:28

Assuming, to begin with, that you don't want bitmap fonts embedded in the PDF, here are some facts about the problem. I'll deal with pdfLaTeX, for XeLaTeX or LuaLaTeX with fontspec it's a different matter.

  1. With the default OT1 encoding, accents are realized by combining two characters, which makes impossible doing "copy-paste".

  2. The font must be available in .pfb (or .pfa) format.

  3. In order to do "copy-paste" from the PDF, the font should also have a correct correspondence between the glyphs and their names.

The link between a TeX font and its Type1 counterpart is provided by the pdftex.map file. When you use the default output encoding and Computer Modern fonts, the relevant line in pdftex.map is

cmr10 CMR10 <cmr10.pfb

The first column is the TeX font name, the second is the PostScript name found in the loaded file, which is cmr10.pfb. Note that when using 11pt type you really are using the scaled 10 point font.

When the T1 output encoding is chosen, instead of the Computer Modern fonts, an extension with accented characters is used: the European Modern fonts. They are not exactly alike CM fonts, but for practical purposes we assume they are. The relevant line in pdftex.map is

ecrm1095 SFRM1095 " T1Encoding ReEncodeFont " <cm-super-t1.enc <sfrm1095.pfb

which is more complex than the other one. The sfrm1095.pfb font file indeed contains a huge number of glyphs: it is used also for the TS1, T2A, T2B, T2C and X2 encodings (text companion and Cyrillic fonts). Thus only a part of it must be picked up, which is done by the ReEncodeFont instruction.

These Type1 counterparts for the European Modern fonts are provided by the so-called CM-Super fonts, that are not included in minimal distributions. So if you want that people can compile the same TeX document with the same result, ensure they have the (meta)package from their TeX distribution.

An alternative is using Latin Modern fonts. When you have a document such as



the Type1 font will be chosen according to the line

ec-lmr10 LMRoman10-Regular " enclmec ReEncodeFont " <lm-ec.enc <lmr10.pfb

Without the fontenc package, the font will be given by

rm-lmr10 LMRoman10-Regular " enclmrm ReEncodeFont " <lm-rm.enc <lmr10.pfb

In the lm-rm.enc file also glyphs in the "upper half" of the font table are defined, but the correspondence is only similar to the Latin-1 encoding.

If you plan to use accented characters in your TeX input file, always add the corresponding call of inputenc and the correct call of fontenc. Otherwise you might get surprising results, as the following MWE shows (note the commented out lines):

% -*- coding: latin-1 -*-






enter image description here

You'd get the same by uncommenting only the fontenc line.

share|improve this answer
Very very good explanation. I never had imagined how it is so complicated to learn about fonts. Well, I don't have ß on my keyboard but I'm able to copy/paste it on my TeXstudio and compile to see the right result on the PDF. I believe that inputenc is working there. It would be the same when I type ç (which is available on my keyboard but not on an English keyboard). – Sigur Mar 17 '13 at 15:28
That's one reason, why XeTeX and LuaTeX were developed, which make life easier here. In XeTeX the simple \documentclass{article} \usepackage{fontspec} \begin{document} dimensões \end{document} will work. – Toscho Mar 17 '13 at 20:12
@Toscho, not for me. xetex foo.tex does not work, produces undefined control sequence. – Sigur Mar 19 '13 at 1:22
I'm sorry. My comment is kind of misleading if one is not accustomed to the difference between TeX Engine and Format: TeX, pdfTeX, XeTeX, LuaTeX, … are engines. plainTeX, LaTeX and ConTeXt are formats, that can work with different engines. My comment document begins with \documentclass{article}, which is only defined with the LaTeX format. So you shouldn't compile the document with xetex foo.tex, which means XeTeX engine with plainTeX format, but with xelatex foo.tex, which means XeTeX engine with LaTeX format. – Toscho Mar 19 '13 at 15:16

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.