I am proficient with TeX but have never used LaTeX and would prefer not to learn it - though perhaps that preference is the sole source of my difficulties.

I am trying to figure out how to produce TeX source that will (a) allow me to include figures in my papers and (b) be acceptable to the arXiv. I can do (a) or (b), but not both at once.

To include figures in my papers, I use the \includegraphics command from LaTeX. To make this possible, I input the file miniltx.tex, a subset of LaTeX written by David Carlisle. I then input the file graphicx.tex (by David Carlisle and Daniel Luecking), which slightly extends miniltx.tex. Now \includegraphics works. Sort of.


  1. If I produce my figures as PDFs, and hand the PDF files to the \includegraphics command, and then run everything through pdftex, I get a PDF file which looks exactly as it should look. For most purposes, this is all I need. However, if I want to submit to the arXiv, they insist on receiving the TeX source. Here's where the trouble starts...

  2. If I produce my figures as PDFs, hand the PDF files to the \includegraphics command, and then run everything through TeX, I get the error "Cannot determine size of graphic in schematic.pdf (no BoundingBox)."

  3. If I produce my figures in any other format (BMP, JPG, etc), hand them to the \includegraphics command, and then run everything through TeX, I get the error message "Unknown graphics extension".

  4. If I produce my figures as anything other than BMPs, give up on miniltx and just use the TeX \special command (I am using MikTeX), TeX appears to run properly, but my previewer chokes on the DVI file, complaining of an "Unknown graphics inclusion type".

  5. If I produce my figures as BMPs, give up on miniltx and just use the TeX \special command, TeX appears to run properly, and the previewer doesn't complain, but the figures are badly distorted and entirely the wrong size.

  6. Since I have no TeX source that can produce a readable DVI file, I want to simply apply pdftex and submit the resulting PDF to the arxiv. The arXiv rejects the paper because it was produced from TeX, which means that I am required to submit the TeX source.

  7. I try to bypass this problem by printing the PDF and scanning it, to produce a new pdf that bears no trace of having come from TeX. The arXiv rejects scanned PDFs.

  8. I try to bypass this problem by converting from PDF to Postscript and then back to PDF again, erasing the telltale traces of TeX along the way. The arXiv insists that this is a "scanned PDF" and rejects it.

So the only alternatives I can see are:

A. Don't submit papers with figures to the arxiv, or

B. Learn LaTeX.

Is there an alternative (C)?

  • 2
    Look at it from another angle (your point B). All TeX commands should work in LaTeX, you can learn a small subset of LaTeX, in less than an afternoon. Can you post a minimal of an example as to what you use at present?
    – yannisl
    Oct 15, 2011 at 6:19
  • Is "C. Use xetex" an option? (i.e., \XeTeXpicfile for images)
    – morbusg
    Oct 15, 2011 at 6:26
  • 2
    One option is to submit only the pdf. Their automatic system will determine that the pdf is geenrated by tex, and therefore reject it. I think that you get an email about that. You can then reply to that email and say that you want to run pdftex but it is not available on their system, so you have no option but to upload a pdf. Most likely, they will then accept the pdf. I have followed this route for papers that I write using ConTeXt.
    – Aditya
    Oct 15, 2011 at 17:00

1 Answer 1


Reading http://arxiv.org/help/submit, I see that that it says

Figures for a submission can be in a number of common formats. These include:

  • PostScript (PS/EPS, most common for line drawings, plots, etc.)
  • JPEG (ideal for photographic images)
  • GIF or PNG (best for non-photographic images, bitmaps)
  • PDF figures (only with PDFLaTeX)


For plain TeX submissions, use the plain tex interface to the graphics package (graphicx.tex) or a macro package like epsf or epsfig

strongly suggesting that PDF graphics can only be used with pdfLaTeX.

The implication is that for plain TeX submissions you have to use the traditional dvips route, and thus EPS graphics. That would explain the messages you get: TeX is running in DVI-output mode, so the graphics package is trying (and failing) to read the bounding box data for your images.

You don't say what the source of your images is, so you may have to convert from PDF to EPS, or may be able to directly generate EPS files.

  • 1
    My understanding is that arXiv doesn't support pdftex, only pdflatex, latex, and tex. Converting the figures to eps is probably the easiest solution, although then you'll likely run into arxiv's length limits.
    – Lev Bishop
    Oct 16, 2011 at 0:47

You must log in to answer this question.

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